View PDF VersionPrevious ArticleNext Article
 
DOI: 10.1039/C5RA06063J (Paper) RSC Adv., 2015, 5, 48484-48491

Studies of membrane fouling mechanisms involved in the micellar-enhanced ultrafiltration using blocking models

Wenxiang Zhangabd, Wenzhong Liangc, Guohe Huangb, Jia Wei*d, Luhui Dinga and Michel Y. Jaffrine
aEA 4297 TIMR, Technological University of Compiegne, 60205 Compiegne Cedex, France
bMOE Key Laboratory of Regional Energy and Environmental Systems Optimization, Resources and Environmental Research Academy, North China Electric Power University, Beijing 102206, China
cSouth China Institute of Environmental Sciences, Ministry of Environmental Protection, Guangzhou 510655, China
dKey Laboratory of Beijing for Water Quality Science and Water Environment Recovery Engineering, Beijing University of Technology, Beijing 100124, China. E-mail: weijia973@gmail.com; Tel: +86-10-67391726
eUMR7338, Technological University of Compiegne, 60205 Compiegne Cedex, France

Received 5th April 2015 , Accepted 18th May 2015

First published on 19th May 2015

Abstract

Micellar-enhanced ultrafiltration (MEUF) is a promising technology to remove organic contaminants from wastewater. A laboratory experiment was carried out to study the performance of four Gemini surfactant systems (CG12, CG16, CG12 + Brij35, CG16 + Brij35) for treating synthetic wastewater. The results show that CG16–Brij35 possesses the lowest flux behavior and the best rejection efficiencies toward phenol and surfactants. Pore blocking models are utilized to estimate membrane fouling of surfactant systems. For CG12 and CG12 + Brij35 systems, internal pore blocking, intermediate pore blocking and cake formation models occur simultaneously. Internal pore blocking and cake formation are the main fouling mechanisms of CG16 micelles. With respect to CG16–Brij35, cake layer is the main type of fouling mechanism. The increased pore blocking degree follows the order of CG12 < CG16 < CG12 + Brij35 < CG16 + Brij35. Furthermore, SEM, ATR-FTIR and a mercury porosimeter are used to analyze membrane surface morphology, membrane material characteristics and membrane fouling layer, as well as examine and verify fouling mechanisms. Due to high flux behavior, excellent phenol rejection and low membrane fouling, CG12 is a more desirable selection. Studies output helps to understand the fouling mechanism of MEUF in the presence of Gemini surfactants, and gave valuable evidence to optimize operation strategies.

1 Introduction

Large amounts of industrial effluents (e.g., antioxidants, manufacture of papers, plastics and dyes) containing phenols present serious adverse effects on human health, most of aquatic life and environment, even at a very low concentration, due to their undesired odors and toxicity.1,2 When they become chlorinated, the impact of phenol molecules will become even more dangerous. Many traditional methods such as UV oxidation, extraction, adsorption, chemical oxidation and biological treatment, have been employed for removing phenol from wastewater.3,4 However, these methods possess inherent deficiencies on account of their low efficiency, high cost, inferior selection and difficulty in removing low molecular weight organic pollutants (e.g. phenols) from wastewater.5 Thus, developing an economical and effective treatment process becomes an urgent problem.4

A promising technology, MEUF, which combines an ultrafiltration membrane and surfactant, has attracted considerable attention.6,7 In MEUF process, surfactants are added to the wastewater with the concentration exceeding a certain value when a great number of micelles is formed.8 These micelles are composed of an outlayer (formed by hydrophilic groups), inner core (formed by hydrophobic groups) and palisade layer (formed by CH2 groups). The organic pollutants tend to dissolve in the palisade layer or the inner core via the rule of similarity.9 This method has been shown to be highly effective in the treatment of pollutants traces, such as heavy metal ions and small organic molecules. In addition, MEUF can be easily integrated with other treatment processes. For instance, the method which combined MEUF and coagulation could improve treatment efficiency for reactive dye wastewater.8,10 Coupling processes of MEUF and electrolysis is excellent for heavy metal wastewater treatment.11,12 However, due to the inherent disadvantages of membrane filtration, flux decline caused by concentration polarization and membrane fouling leads to the reduction of operation efficiency in MEUF, which limits the success of its industrial application.13,14 Moreover, the complex fouling mechanism of MEUF and its behavior have not yet been studied. In order to control flux decline of MEUF, it is important to investigate fouling mechanisms and determine the most appropriate operational strategies and membrane restoration procedures.

In previous investigations, a number of mathematical models were utilized to explain membrane fouling mechanisms, including cake filtration models,15 concentration polarization models16 and blocking models.17,18 Concentration polarization models are suitable in the case of fouling dominated by concentration polarization, while cake filtration models can be used to describe particles accumulation on the membrane surface. The membrane blocking model is appropriate to analyze organic fouling: when the size of membrane pores is larger than that of particle, solutes may penetrate and block the membrane pores and a cake layer forms on membrane surface. Based on pore-blocking laws, four different kinds of blocking models (complete blocking, intermediate blocking, standard blocking, and cake formation) proposed by Hermia et al.19 are used to explain the fouling of dead-end filtration. The four blocking models are presented in Table 1. Hwang et al.20 claimed that various membranes had different blocking types, e.g., standard blocking occurred in MF-Millipore membrane, while a complete blocking existed in an isopore membrane and a durapore membrane presented cases of intermediate blocking. Besides, during filtration process, the blocking models might change from a complete blocking to cake formation model. Ho et al.21 put forward a new mathematical model to calculate the thickness of cake layer formed on the membrane surface in protein filtration. In order to simplify the cumbersome process that applies different models to explain fouling in the entire filtration process, another smooth transition model was proposed to explain the change from pore blocking fouling to cake filtration resistance with time.22 Recently, some studies23,24 used a blocking model to explain the fouling mechanism of cross-flow filtration. In cross-flow filtration, the amount of solutes contained in the convective flow along membrane surface is larger than that removed by cross flow action and similar to dead-end filtration. Furthermore, to our knowledge, there is no available description of the membrane fouling mechanisms of MEUF by pore blocking models.

Table 1 Four models of membrane fouling proposed by Hermia23
Pore blocking models Hermia's model Physical concept Schematic diagram
Complete pore blocking (model 1) J = J0 × exp(−K × t); K (s−1) Formation of a surface deposit image file: c5ra06063j-u4.tif
Internal pore blocking (model 2) J = (J0−0.5 + K × t)−2; K (m s)−0.5 Pore blocking + surface deposit image file: c5ra06063j-u5.tif
Intermediate pore blocking (model 3) J = (J0−1 + K × t)−1; K (m−1) Pore constriction image file: c5ra06063j-u6.tif
Cake formation (model 4) J = (J0−2 + K × t)−0.5; K (m s)−1 Pore blocking image file: c5ra06063j-u7.tif

In this paper, the effect of various Gemini surfactant systems (CG) on fouling type and blocking index are investigated. Gemini surfactant is a promising surfactant including two hydrophilic head groups and two hydrophobic chains which are connected at or near the head groups by a relatively short spacer group. Moreover, blocking models integrated with an exhaustive statistical analysis are used to analyze fouling mechanism in MEUF for phenol wastewater treatment. At the same time, the treatment efficiency and flux behavior are studied. Scanning electron microscopy (SEM), Fourier transform infrared spectrometer with attenuated total reflectance accessory (ATR-FTIR) and mercury porosimeter are used to examine and verify fouling mechanisms. This study is expected to facilitate their potential application to continuous filtration tests of MEUF in the future.

2 Theory

Based on the pore blocking mechanism, the permeate flow decline can be described by
 
image file: c5ra06063j-t1.tif(1)
where V and t are the permeate volume and filtration time, and k is the blocking index, where n is a dimensionless number of the fouling mechanism.15,16,25 Normally, four pore blocking fouling mechanisms (listed in Table 1) are considered:

In complete pore blocking, all particles depositing on membrane surface participate in “sealing” of membrane pores. It is an idealized condition assuming that no particles are located on top of other particles or on membrane surface. Complete pore blocking corresponds to n = 2 with the equation:

 
image file: c5ra06063j-t2.tif(2)
where kb is a complete pore blocking coefficient, Am is the membrane area, J0 is the initial flux and Jn=2 is the flux.

Internal pore blocking is due to the constriction of membrane pores and small particles attach into pore walls. Internal pore blocking (n = 1.5) is described by:

 
image file: c5ra06063j-t3.tif(3)
where ks is an internal pore blocking coefficient and Jn=1.5 is the flux.

For intermediate pore blocking, particles depositing on membrane not only cause pore blocking, but also attach each other on membrane surface. Intermediate pore blocking (n = 1):

 
image file: c5ra06063j-t4.tif(4)
where ki is an intermediate pore blocking coefficient and Jn=1 is the flux.

In the case of cake formation, particles don't cause any changes in membrane pores. A cake layer forms outside the external membrane and increases hydraulic resistance. In that case n = 0:

 
image file: c5ra06063j-t5.tif(5)
where kc is a cake pore blocking coefficient and Jn=0 is the flux.

3 Materials and methods

3.1 Materials

Two kinds of Gemini surfactants (referred to as CG12 and CG16) were supplied by Chengdu Organic Chemicals Co. Ltd, Chinese Academy of Science, with a purity of 98%. The nonionic surfactant (Brij35) with purity 98% was obtained from Sigma-Aldrich. The molecular structures and properties of selected surfactants are given in Table 2. Phenol with analysed purity was purchased from the Beijing Chemical Reagent Company, China. All reagents were used without further purification. Distilled water was used for solution preparation in all experiments.
Table 2 The physicochemical properties of surfactants and phenol in experiment
Surfactant Structure MW (g mol−1) CMCa (mM) Sw (20 °C) log Kow
a Error limits of CMCs are ±4%.
Phenol image file: c5ra06063j-u8.tif 94.11   8.3 g L−1 1.46 (ref. 9)
CG12 image file: c5ra06063j-u9.tif 614.67 0.8 (ref. 29)    
CG16 image file: c5ra06063j-u10.tif 672.03 0.2 (ref. 32)    

3.2 Methods

3.2.1 MEUF measurements. MEUF was conducted at 20 °C with a flat sheet module built by Xiamen Tianquanxin Membrane Technology Co, Ltd, China. The flat sheet membranes used in this study were obtained from Advanced Membrane Corporation, America. The membrane material (10 kDa, 0.06 m2) was hydrophobic polyethersulfone (PES).

MEUF measurements were carried out at 20 °C using a flat sheet module supplied by Xiamen Tianquanxin Membrane Technology Co, Ltd., China, which was presented in our previous study.26 The feed wastewater was made by adding the pre-determined amount of phenol and surfactant into deionized water and then was mixed adequately to make sure that the solutes evenly dispersed before entering the membrane module. The retentate flow rates and pressures were kept constant at 5 L min−1 and 0.30 MPa, with a 4.0 L initial volume of feed solution. The power, flow rate and pump head were 1.1 kW, 2 m3 h−1 and 40 m, respectively. The experiments were carried out with the feed and retentate solution being recycled back to the feed tank. The process was stopped after 420 min of operation. The permeate fluxes (Jv) were recorded and the permeate solution and retentate solution were collected to analyze their concentrations. All the experiments were repeated three times, and all the deviations were less than 5%. Reported values were the average of three duplicate records.

Thereafter, deionized water was filtered at a pressure of 0.25 MPa to wash the exterior of membrane within 15 min, then 0.1 M NaOH, deionized water, 0.1 M HNO3 and finally deionized water were used to rinse out the membrane at 0.20 MPa for 10 min. After then, deionized water was used to measure the permeate flux for checking the permeability of the membrane.

3.2.2 Analysis. In the synthetic solution, the concentration of phenol was measured by UV-2102 PCS spectrophotometer at 270 nm. The concentration of Gemini surfactants was measured using a titrating method.27 The viscosities of surfactant solutions were analyzed by a viscometer (NDJ-5S/8S).

With the purpose of characterizing the microscopic effects by micelles on membrane fouling, SEM (FEI QUANTA 200, FEI Company, USA), ATR-FTIR (Nicolet 560, Thermo Electron Corporation, USA) and mercury porosimeter (AUTOPORE II 9220, Micromeritics, USA) were applied in this study. The selected membrane, consisting of three layers, inner support layer, outer active layer and intermediate porous connection layer, was cut to separate the flat-sheet active layer alone for the measurements of mercury porosimeter and ATR-FTIR. Before analysis, membrane samples were dried by vacuum under 50 °C for 8 h to examine the moisture.

3.3 Fouling mechanism evaluation

In this study, the fouling type is determined by Hermia blocking model23 (Table 1). The estimation of K-constant associated with each of the models is calculated by nonlinear regression of the experimental data. The F-value test with successive iterations was applied until the sum of the squares' residuals was lower than a standard value.

4 Results and discussion

4.1 Flux behavior

The flux behavior of various surfactant systems is shown in Fig. 1. As expected, for all surfactant systems, a noticeable flux decline occurs before 200 min. However, after 200 min, flux decreases slowly and becomes stable. Therefore, as for the flux decline in MEUF, two stages can be distinguished. At the first stage (t ≤ 200 min), the micelles can deposit or be adsorbed on membrane surface promptly, so the concentration layer and membrane fouling form rapidly and the flux reduces markedly. During the second stage (t ≤ 200 min), the flux reaches a stable value and membrane fouling doesn't increase much, which corresponds to an equilibrium between foulants accumulation and diffusion. According to the interaction effect between membrane and foulants,3 the type of micelles-membrane fouling can be divided into foulant–clean membrane interaction and foulant-deposited-foulant interaction. The membrane fouling type changes from foulant–clean membrane interaction to foulant-deposited-foulant interaction with filtration time. The first stage is mainly controlled by foulant–clean-membrane interaction, thus its flux significantly depends on membrane properties and feed components (pore size, materials, micelles characteristic etc.), while the second stage is independent of membrane and feed components (foulant-deposited-foulant interaction).
image file: c5ra06063j-f1.tif
Fig. 1 Flux behavior with various CG micelle systems.

Fig. 1 shows that optimal flux increased in the order as follows: CG12 > CG16 > CG12 + Brij35 > CG16 + Brij35. Compared with CG16 systems (CG16 and CG16 + Brij35), CG12 systems (CG12 and CG12 + Brij35) are more easily ionized and have more powerful positive charge in solution, respectively, due to the shorter hydrophobic tail,28 thus stronger electrostatic repulsive force may occur between CG12 systems and PES membrane (positive charge), reducing concentrate polarization and membrane fouling. Besides, because of the lower molecular weight and the higher hydrophilicity,8 CG12 systems have lower particles accumulation on the membrane surface. With respect to the single surfactant systems (CG12 and CG16), the mixed surfactant systems (CG12 + Brij35 and CG16 + Brij35) show a lower flux and more severe flux decline, which is due to three facts. First, the viscosities of solutions increase with the increment of nonionic surfactant, so the hydraulic resistance against flux enhances; secondly, the molecular weight of mixed surfactant systems is higher than that of single surfactant systems, thus micelles of mixed surfactant systems more easily deposit on membrane surface or block membrane pores; third, nonionic surfactant has a higher adsorption on hydrophobic membrane than ionic surfactants,29 thus the addition of ionic increases the adsorption effect and led to more serious membrane fouling.

4.2 Fouling mechanism

Four blocking models were used to describe the fouling mechanisms of MEUF. Table 3 shows the calculated K-values of the fouling constant, R-square values adjusted for degrees of freedom, the standard error of the estimate and F-value tests. These values are calculated by models described in the equations shown in Table 1 and experimental data in Fig. 1. For each regression analysis, an ANOVA analysis is performed to estimate significances of the model. For all CG systems, the R-squared values of model 4 are higher than 0.992, indicating that all the studied CG systems agree with cake form fouling mechanism and it is useful for describing the phenomenon. Model 1 has lowest R-squared values and complete pore blocking is not suitable for illustrating fouling mechanism of MEUF. With respect to F-value tests, all studied models have much lower values than F-value, therefore there are no significant differences between experimental data and model. Besides, model 4 has lowest residual values, implying that cake formation fouling mechanism has the best fitting degree, while residual values of model 1 are highest and complete pore blocking presents the lowest fitting degree.
Table 3 K-fouling constant values and statistics analysis for fouling mechanism model
UF-conditions Model R-squared (adjusted for degrees of freedom) K-value K-units Residual sum of squares Residual mean square F value Prob > F
CG12 1 0.945 2.51 × 10−3 s−1 0.179 0.016 207.2 1.76 × 10−8
2 0.998 2.46 × 10−4 (m s)−0.5 1.1 × 10−3 9.99 × 10−5 5260.7 4.44 × 10−16
3 0.996 9.77 × 10−5 m−1 9.76 × 10−5 8.88 × 10−6 2764.3 1.45 × 10−14
4 0.997 8.04 × 10−6 (m s)−1 1.64 × 10−7 1.49 × 10−8 4223.2 1.44 × 10−15
CG16 1 0.862 4.12 × 10−3 s−1 1.33 0.12 75.7 2.91 × 10−6
2 0.994 3.72 × 10−4 (m s)−0.5 4.06 × 10−3 3.69 × 10−4 2147.6 5.78 × 10−14
3 0.989 1.77 × 10−4 m−1 5.56 × 10−4 5.06 × 10−5 1128.7 1.94 × 10−12
4 0.995 2.13 × 10−5 (m s)−1 1.58 × 10−6 1.44 × 10−7 2500.8 2.51 × 10−14
CG12–Brij35 1 0.897 2.38 × 10−3 s−1 0.12 0.012 96.8 1.84 × 10−6
2 0.995 4.23 × 10−4 (m s)−0.5 4.07 × 10−3 3.7 × 10−4 2587.5 2.09 × 10−14
3 0.993 2.22 × 10−4 m−1 5.44 × 10−4 4.95 × 10−5 1700.2 2.08 × 10−13
4 0.998 3.26 × 10−5 (m s)−1 1.24 × 10−6 1.13 × 10−7 7191.6 1.11 × 10−16
CG16–Brij35 1 0.807 6.13 × 10−3 s−1 4.33 0.394 51.2 1.85 × 10−5
2 0.987 6.04 × 10−4 (m s)−0.5 0.016 1.42 × 10−3 931.3 5.54 × 10−12
3 0.978 3.59 × 10−4 m−1 3.4 × 10−3 3.05 × 10−4 539.36 1.07 × 10−10
4 0.992 6.94 × 10−5 (m s)−1 2.36 × 10−5 2.14 × 10−6 1494.3 4.2 × 10−13

In CG12 systems, model 2, 3 and 4 have a very high and similar R-squared value (higher than 0.993), so internal pore blocking, intermediate pore blocking and cake formation models occur simultaneously. During this fouling process, at the beginning, the intermediate pore blocking is dominant: some small CG12 or CG12 + Brij35 micelles or monomers enter into the membrane pores, attach into pore wall and narrow membrane pores. Then, more micelles deposit on membrane surface, get involved in “sealing” of membrane pores and membrane pores are further narrowed, to produce internal pore blocking fouling. After that, most membrane pores are blocked and micelles don't participate in any changes of membrane pores, as well as a cake layer composed by micelles forms outside the external membrane and increases hydraulic resistance, due to cake forming. For CG16 micelles, the main fouling mechanisms are internal pore blocking and cake formation. Due to their larger size, CG16 micelles directly deposit on membrane and seal membrane pores, instead of attaching into pore walls, thus intermediate pore blocking is not the main fouling mechanism. With respect to CG16–Brij35 micelles, their size is evidently higher than PES 10 kDa, thus they directly form a cake layer on membrane surface, without attaching into pore walls and “sealing” membrane pores, therefore the main fouling mechanism of CG16–Brij35 is cake formation.

In Hermia's blocking model, a K-value can be used to estimate the degree of membrane pore blocking.23 As shown in Table 3, for internal pore blocking, intermediate pore blocking and cake formation model, the increased K-values have the following order: CG12 < CG16 < CG12 + Brij35 < CG16 + Brij35, which indicates that the thickness of cake layer occurring on membrane surface presents the same sequence as flux behavior in Section 4.1. According to the analysis of Tang et al.,30 foulants characteristic is an important factor influencing fouling mechanism. In this study, fouling mechanism varies with micelle types. The order of the micelle size is CG12 < CG 16 < CG12–Brij35 < CG16–Brij35. The increase of micelle size caused by the addition of surfactant or longer carbon chain enhances the importance of cake formation fouling and improves the degree of pore blocking.

4.3 Phenol and surfactant rejection

Phenol and surfactant rejections with various CG systems are presented in Fig. 2. With the addition of Brij-35, phenol and surfactant rejections increase, which can be explained as follows. In the single CG systems, the formation of micelles is hindered by the electrostatic repulsive forces between the hydrophilic head groups of monomeric CG molecules. When a nonionic surfactant penetrates the micelle formation, the charge of ionic hydrophilic groups is counterbalanced by part of nonionic surfactants.1 Meanwhile, the electrical repulsion in the Stern layer of the micelles trails off. Moreover, the electrical potential reduces due to the diminution of the charge density on the surface of micelles.13 As a consequence, lowered electrical potential excites the formation of micelles with lower CMC, then more CG becomes available in the micellar phase to dissolve phenol. In addition, attachment of ethylene oxygen group (C2H4O) of Brij35 to CG micelles increases the size of the micelles, thus more phenol molecules are dissolved and larger micelles are easily rejected, improving phenol and surfactant rejections. Compared to CG12, owing to longer alkyl chain length and higher hydrophobic, CG16 has a larger size and a stronger combined effect with an ethylene oxygen group (C2H4O) of nonionic surfactant Brij-35, leading to greater stability and capacity of dissolving phenol and enhancing micelles rejection.
image file: c5ra06063j-f2.tif
Fig. 2 Phenol and surfactant rejection with various CG systems.

4.4 Characterization of membrane surface morphology and properties altered by micelles for all selected surfactant systems

For the sake of investigating and verifying fouling mechanism models, the membrane surface morphology, membrane material characteristics and membrane fouling microscopic will be further characterized by several analysis techniques. SEM images are used to observe foulants on the membrane surface. Porosity measurements indicate pore blocking fouling. ATR-FTIR analyses indicate the composition and structure of foulants.
4.4.1 Membrane surface morphology. As shown in Fig. 3, the new membrane is dark, whereas other membranes show white spots indicating they were fouled by all surfactant micelles studied herein. There are widespread cake layers on membranes fouled by micelles, which confirms that a cake formation model can explain the fouling mechanism of MEUF. Obviously, cake layers are thicker in images (c) and (d) demonstrating the more serious membrane fouling for micelles of CG12–Brij35 (c) and CG16–Brij35 (d). This can be attributed to: the mixed surfactant systems more easily form cake layers on membrane surface. Additionally, comparing with the membranes of CG12 (a) and CG12–Brij35 (c), the membranes of CG16 (b) and CG16–Brij35 (d) have thicker cake layers, respectively. This is because the higher micelle molecular weight of CG16 leads to a larger accumulation of CG16 micelles on membrane surface. These results are consistent with those discussed in Section 4.2.
image file: c5ra06063j-f3.tif
Fig. 3 SEM images showing the surface morphologies (20[thin space (1/6-em)]000×) of (a) fouled membrane of CG12, (b) fouled membrane of CG16, (c) fouled membrane of CG12–Brij35 and (d) fouled membrane of CG16–Brij35.
4.4.2 Membrane material characteristics. Fig. 4 presents the changes of porosity in terms of pore volume per unit mass of membrane material with pore size. The distribution of pore diameters of fouled membranes is narrower than that of new membrane to a certain extent, due to the existence of pore blocking fouling and cake layers.31 The sequence of the porosity of the membranes is: new membrane > fouled membrane of CG12 > fouled membrane of CG16 > fouled membrane of CG12–Brij35 > fouled membrane of CG16–Brij35, which coincides well with the sequence of pore diameters of the membranes. These are also in agreement with the degree of pore blocking and cake layer. The single surfactant systems have a larger volume per unit mass of membrane material than mixed surfactant systems, probably owing to their lower micelle molecular weights and smaller viscosities. Furthermore, CG12 systems have higher porosities than CG16 systems, which corresponds with the analysis of Section 4.2. When micelles deposit on membrane surface, smaller pores may be easily blocked by the smaller size of micelles. Although most of the small pores are blocked, the porosity of smaller pores increases rather than decreases. This may be attributed to some large pores that are partially blocked, thus micelles accumulate in small pores.
image file: c5ra06063j-f4.tif
Fig. 4 Porosity changes of new and fouled membrane.
4.4.3 Analysis of membrane foulants. The FTIR patterns of the new and fouled membranes are shown in Fig. 5. It is observed that the peaks of all membrane material (image file: c5ra06063j-u1.tif) can be assigned as follows: 837/872 cm−1, aromatic hydrocarbons 1, 4-replacement image file: c5ra06063j-u2.tif; 1149/1298/1321 cm−1, R–SO2–R′; 1485/1578/3070 cm−1, image file: c5ra06063j-u3.tif and 1660 cm−1, aromatic hydrocarbon compounds.31,32 Compared with the new membrane, the peaks of fouled membrane at 2856/2923 cm−1 (CH2) provide evidences of the existence of alkyl-like substances in the cake layer, which contains micelles. The strengthened ATR-FTIR spectra of mixed surfactant systems demonstrate a number of micelles deposits on the membrane surface or enter the pores of membrane. Based on the ATR-FTIR results, cake layers of mixed surfactant systems are thicker than single surfactant systems, which agrees with their higher pore blocking.
image file: c5ra06063j-f5.tif
Fig. 5 ATR-FTIR spectrums of (a) new membrane, (b) fouled membrane of CG12, (c) fouled membrane of CG16, (d) fouled membrane of CG12 + Brij35 and (e) fouled membrane of CG16 + Brij35.

5 Discussion

As a promising membrane wastewater treatment technology, MEUF presents excellent removal efficiency for trace amounts of organic pollutants. However, there is not yet a successful practical engineering example. Membrane fouling caused by organic matters (micelles) is an important drawback. This organic fouling is mainly due to pore blocking and cake formation on membrane and micelles play a significant role on the fouling types and degree of pore blocking. Due to longer carbon chain and higher hydrophobic, CG12 systems have a lower degree of pore blocking than CG16 systems. Because of larger molecular weight, bigger viscosities and weaker electrostatic repulsive force between micelles and membrane, mixed surfactant systems induce a higher fouling than single surfactant systems. In addition, three advanced analysis techniques were used to verify to actual fouling degree for various micelles. Regarding membrane cleaning, the fouled membrane with lower degree of pore blocking and cake layer recovers more easily its permeability, thus choosing a suitable surfactant system is important to maintain long-term reuse of membrane and simplify cleaning process. Among these four surfactant systems, CG12 has a much higher flux than others (ca. 50–300%) and its phenol rejection is 87% (the highest value is 95%). Compared with the similar phenol rejection, flux behavior as a critical parameter of process efficiency is more important. Besides, CG12 has a lower membrane fouling (Section 4.2 and 4.4). So we think that CG12 is a more desired selection (Fig. 6). With the proper membrane plant design, a suitable surfactant system can keep a steady and sustainable operation and high productivity and the longevity of the membrane modules can be extended to a maximum.
image file: c5ra06063j-f6.tif
Fig. 6 Membrane fouling mechanism of MEUF with various micelle systems.

6 Conclusions

In this study, MEUF assisted by four CG systems was used to treat phenol wastewater. Flux behavior, fouling mechanism and separation performance were investigated. CG16–Brij35 has the best rejection efficiencies for phenol and surfactant as well as the lowest permeate flux owing to largest molecular weight, biggest viscosities and weakest electrostatic repulsive force Moreover, pore blocking models proposed by Hermia are utilized to estimate membrane fouling of four CG systems in MEUF. As for CG12 systems, internal pore blocking, intermediate pore blocking and cake formation are the main fouling mechanisms. The fouling mechanisms of CG16 are internal pore blocking and cake formation, while cake formation is the main fouling mechanism of CG16–Brij35. The change of fouling mechanism depends on micelles characteristics. Besides, the increased pore blocking degree follows the order: CG12 < CG16 < CG12 + Brij35 < CG16 + Brij35. In addition, SEM, ATR-FTIR and a mercury porosimeter were used to analyze membrane surface morphology, membrane material characteristics and membrane fouling for the fouled membranes, as well as examine and verify fouling mechanisms. Cake layer is the main form of membrane fouling in MEUF and the thickness of cake layer is consistent with pore blocking degree. With respect to other surfactant systems, CG12 has high flux behavior, excellent phenol rejection and low membrane fouling and is the best choice. These findings not only clarify the fouling mechanism of MEUF, but also provide valuable advice for fouling control.

Acknowledgements

This research is supported by the National Natural Science Foundation (51209088) and the Natural Science and Engineering Research Council of China. Besides, the financial support of China Scholarship Council (CSC) for Wenxiang Zhang's thesis fellowship is acknowledged.

References

  1. J. Lee, J.-S. Yang, H.-J. Kim, K. Baek and J.-W. Yang, Desalination, 2005, 184, 395–407 CrossRef CAS PubMed.
  2. J. Luo, F. Marpani, R. Brites, L. Frederiksen, A. S. Meyer, G. Jonsson and M. Pinelo, J. Membr. Sci., 2014, 459, 1–11 CrossRef CAS PubMed.
  3. J. Luo, W. Cao, L. Ding, Z. Zhu, Y. Wan and M. Y. Jaffrin, Sep. Purif. Technol., 2012, 96, 194–203 CrossRef CAS PubMed.
  4. S. Jadhav, N. Verma, A. Sharma and P. Bhattacharya, Sep. Purif. Technol., 2001, 24, 541–557 CrossRef CAS.
  5. M. M. Husein, A. Deriszadeh and T. G. Harding, Desalination, 2011, 273, 352–358 CrossRef CAS PubMed.
  6. W. Zhang, G. Huang, J. Wei and D. Yan, Desalination, 2013, 311, 31–36 CrossRef CAS PubMed.
  7. H. A. Namaghi and S. M. Mousavi, J. Taiwan Inst. Chem. Eng., 2014, 45, 1850–1854 CrossRef CAS PubMed.
  8. P. Yenphan, A. Chanachai and R. Jiraratananon, Desalination, 2010, 253, 30–37 CrossRef CAS PubMed.
  9. W. Zhang, G. Huang, J. Wei, H. Li, R. Zheng and Y. Zhou, J. Hazard. Mater., 2012, 235, 128–137 Search PubMed.
  10. A. Ahmad and S. Puasa, Chem. Eng. J., 2007, 132, 257–265 CrossRef CAS PubMed.
  11. F. Luo, G.-M. Zeng, J.-H. Huang, C. Zhang, Y.-Y. Fang, Y.-H. Qu, X. Li, D. Lin and C.-F. Zhou, J. Hazard. Mater., 2010, 173, 455–461 CrossRef CAS PubMed.
  12. C.-K. Liu and C.-W. Li, Sep. Purif. Technol., 2005, 43, 25–31 CrossRef CAS PubMed.
  13. E. Zeftawy, M. Monem and C. N. Mulligan, Sep. Purif. Technol., 2011, 77, 120–127 CrossRef PubMed.
  14. B. Rahmanian, M. Pakizeh, S. A. A. Mansoori, M. Esfandyari, D. Jafari, H. Maddah and A. Maskooki, J. Taiwan Inst. Chem. Eng., 2012, 43, 558–565 CrossRef CAS PubMed.
  15. A. Grenier, M. Meireles, P. Aimar and P. Carvin, Chem. Eng. Res. Des., 2008, 86, 1281–1293 CrossRef CAS PubMed.
  16. H. Huang, T. A. Young and J. G. Jacangelo, Environ. Sci. Technol., 2007, 42, 714–720 CrossRef.
  17. K.-J. Hwang, C.-Y. Liao and K.-L. Tung, J. Membr. Sci., 2007, 287, 287–293 CrossRef CAS PubMed.
  18. K.-J. Hwang and C.-Y. Liao, J. Taiwan Inst. Chem. Eng., 2012, 43, 46–52 CrossRef CAS PubMed.
  19. P. Hermia and H. Bredée, J. Soc. Chem. Ind., 1936, 55, 1–4 CrossRef PubMed.
  20. K.-J. Hwang and T.-T. Lin, J. Membr. Sci., 2002, 199, 41–52 CrossRef CAS.
  21. C.-C. Ho and A. L. Zydney, J. Colloid Interface Sci., 2000, 232, 389–399 CrossRef CAS PubMed.
  22. H. Rezaei, F. Z. Ashtiani and A. Fouladitajar, Desalination, 2011, 274, 262–271 CrossRef CAS PubMed.
  23. B. Cancino-Madariaga, R. Ruby, C. Astudillo Castro, J. Saavedra Torrico and M. Lutz Riquelme, Ind. Eng. Chem. Res., 2012, 51, 4017–4024 CrossRef CAS.
  24. M. Palencia, B. L. Rivas and H. Valle, J. Membr. Sci., 2014, 455, 7–14 CrossRef CAS PubMed.
  25. S. Chellam and W. Xu, J. Colloid Interface Sci., 2006, 301, 248–257 CrossRef CAS PubMed.
  26. W. Zhang, G. Huang and J. Wei, Desalin. Water Treat., 2014, 1–11 Search PubMed.
  27. Y. Sun, Y. Feng, H. Dong and Z. Chen, Surf. Sci., 2007, 601, 1988–1995 CrossRef CAS PubMed.
  28. J. Wei, G. Huang, L. Zhu, S. Zhao, C. An and Y. Fan, Chemosphere, 2012, 89, 1347–1353 CrossRef CAS PubMed.
  29. A.-S. Jönsson, B. Jönsson and H. Byhlin, J. Colloid Interface Sci., 2006, 304, 191–199 CrossRef PubMed.
  30. C. Y. Tang, T. Chong and A. G. Fane, Adv. Colloid Interface Sci., 2011, 164, 126–143 CrossRef CAS PubMed.
  31. M. Aoudia, N. Allal, A. Djennet and L. Toumi, J. Membr. Sci., 2003, 217, 181–192 CrossRef CAS.
  32. J. Wei, G. Huang, H. Yu and C. An, Chem. Eng. J., 2011, 168, 201–207 CrossRef CAS PubMed.
This journal is © The Royal Society of Chemistry 2015
Click here to see how this site uses Cookies. View our privacy policy here.