The use of micellar-enhanced ultrafiltration (MEUF) for fluoride removal from aqueous solutions

Accepted Manuscript The use of micellar-enhanced ultrafiltration (MEUF) for fluoride removal from aqueous solutions Martyna Grzegorzek, Katarzyna Majewska-Nowak PII: DOI: Reference:

S1383-5866(17)32369-9 https://doi.org/10.1016/j.seppur.2017.11.022 SEPPUR 14178

To appear in:

Separation and Purification Technology

Received Date: Revised Date: Accepted Date:

29 June 2017 3 November 2017 10 November 2017

Please cite this article as: M. Grzegorzek, K. Majewska-Nowak, The use of micellar-enhanced ultrafiltration (MEUF) for fluoride removal from aqueous solutions, Separation and Purification Technology (2017), doi: https:// doi.org/10.1016/j.seppur.2017.11.022

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

The use of micellar-enhanced ultrafiltration (MEUF) for fluoride removal from aqueous solutions Martyna Grzegorzek1,*, Katarzyna Majewska-Nowak2 1,2

Wroclaw University of Science and Technology, Department of Environmental Engineering, WybrzeżeWyspiańskiego 27, 50-370 Wroclaw, Poland *

Corresponding author e-mail: [email protected] 2 e-mail: [email protected]

Keywords: fluoride, micellar ultrafiltration, CPC, ODA, membranes ABSTRACT Fluorine is a common chemical element. Due to the harmful influence on human health, the World Health Organization determined the permissible level of fluoride in drinking water to be 1.5 mg F-/L. Fluorine can be removed from water solutions by various methods (i.e. ion-exchange, precipitation, adsorption and membrane techniques). One membrane technique that can be effective in fluoride removal is micellar enhanced ultrafiltration (MEUF). This method involves the ability of surfactants to create micelles, which are retained by classic ultrafiltration membranes. In the presented paper, the suitability of the MEUF process for fluoride removal was verified. Polyethersulfone and regenerated cellulose ultrafiltration membranes, as well as two cationic surfactants (cetylpyridinium chloride CPC, octadecylamine acetate - ODA),were used in the batch experiments. The fluoride content in the model solutions amounted to 10 and 100 mg F-/L, whereas surfactant concentration varied in the range of 1-3 CMC. The MEUF tests were performed under a pressure of 0.2 MPa with and without a salt (NaCl) dosage to the treated solutions. The results obtained showed that for a low fluoride content (10 mg F-/L) and a high CMC value (3CMC), the polyethersulfone membrane allowed F- ions below the permissible level for drinking water to be removed. The presence of NaCl in the model solutions resulted in a significant worsening of fluoride removal efficiency.

1. INTRODUCTION Fluorine is a common chemical element present in the natural environment. Due to the harmful impact on human health, the World Health Organisation (WHO) determined the permissible limit of fluorine in drinking water as 1.5 mg F-/L [1–3]. A low content of fluoride has a beneficial influence on a human's organism. An excess of fluoride may lead to fluorosis, brain damage or dental caries [1– 4].The problem of elevated fluoride concentration in water is present all over the world [5–6]. In Tanzania, the fluoride content can exceed 330 mg F-/L, whereas a concentration of 180 mg F-/L is possible in the groundwater resources of Kenya [5]. This problem was also discovered in Pakistan, USA, Canada, China and Thailand [6]. There is a need to find new ways of overcoming the problem of water contamination by fluorine. Various methods can be used for fluoride removal such as: ionexchange, adsorption, precipitation and membrane processes (i.e. electrodialysis, membrane distillation, nanofiltration and reverse osmosis) [2,7]. One possible way to solve the problem of elevated fluoride concentration may be micellar enhanced ultrafiltration (MEUF). The first tests with MEUF were made by Dunn et al. [8] in the mid80s. This method is usually used for organic compounds, inorganic substances or the removal of metal ions from water. It is especially suitable for the removal of ionic and hydrophobic components. The MEUF process is based on conventional ultrafiltration (UF) and surfactants, which have a hydrophobic tail and hydrophilic head. Micelles lead to surface tension reduction and can contain even 150 molecules. Surfactants can be divided into three types: cationic, anionic and non-ionic. When surfactant concentration is greater than the critical micelle concentration (CMC), micelles are created. Contaminants are attached to the external surface of the micelle by electrostatic interactions or are

retained in the hydrophobic core of micelles. Each micelle has a surface charge and is able to bind an ion with the opposite charge. In the next step, the created micelles can be separated during ultrafiltration because their size is larger than the diameter of membrane pores [9–17]. The MEUF process has many advantages, such as low energy costs, high removal efficiency and high fluxes or pollutant recovery. Moreover, it is less energy intensive in comparison with conventional purification methods. Micellar enhanced ultrafiltration can also be used for the recovery of valuable compounds. However, it can also be seen that one of the problems related to the MEUF process is fouling and concentration polarization. Membrane fouling leads to a shortening of the membrane lifetime and it decreases membrane efficiency. The most common reasons of membrane fouling are: compression of the gel layer, pore blocking by contaminants, concentration polarization, and also adsorption of solids onto the membrane surface and in the membrane matrix. In the case of concentration polarization, micelles or other contaminants are deposited on the surface of the membrane. As a result, a fouling layer is formed, which has a negative influence on the hydraulic properties of the membrane - it decreases the permeate flux. Moreover, it should be noted that small amounts of surfactant monomers, as well as unfixed pollutant molecules, can be transported through the membrane to the permeate. This undesirable phenomenon has a negative impact on permeate quality. The high cost of surfactants could also be a problem, however, there is a possibility of surfactant recovery [18]. In the MEUF process, the main factors influencing the removal efficiency and permeate flux are the operating conditions, membrane and surfactant properties and also the characteristics of the solution [19-25]. In the course of the MEUF process, mechanical filtration plays the main role. However, the MEUF process is also governed by other phenomena, such as adsorption of micelles, solubilisation and micellization [26]. The created micelles of cationic surfactants are characterized by a positive surface electrical charge and they can interact electrostatically with anionic type pollutants. The MEUF process is conducted by applying cationic surfactants and can be used in the removal of anionic contaminates, such as anionic heavy metals [20, 27, 28], anionic organic dyes [29], phosphates [30] or nitrates [31]. Among cationic surfactants, the most common and effective in the removal of anionic pollutants by MEUF is cetylpyridinium chloride (CPC). Iqbal et al. [27] investigated the separation of arsenate using various cationic surfactants and revealed that the highest removal efficiency (96%) was obtained for CPC. With the help of this surfactant it is possible to remove 99% of chromate ions [28], however, the rejection rate is strongly influenced by the ionic strength and pH solution. A high NaCl concentration (0.5 mol/L) in a chromate-surfactant solution caused a considerable drop in chromate ions removal (to 46%). It has recently been reported that the MEUF process was successfully applied in the removal of emerging contaminants. Acero et al. [11] almost achieved a complete removal of diclofenac and ketorolac, and 95% retention of sulfamethoxazole with the use of CPC at a concentration of 2 CMC. The research demonstrated a strong correlation between micellar binding and the pKa values of selected compounds, indicating that electrostatic interactions between anionic compounds and cationic surfactants play an important role in MEUF. Some attempts have been made towards nutrient removal using the MEUF process. Baek et al. [31] investigated the MEUF process with the use of CPC for nitrate and phosphate removal and found that surfactant to pollutant molar ratios influenced separation efficiency. 86% of nitrate and 91% of phosphate were removed when the molar ratio of CPC to total pollutants was higher than 3. Camarillo et al. [30] applied CPC, as well as octadecylamine acetate (ODA), to decrease the phosphate concentration in treated domestic wastewater. The phosphate rejection amounted to 95% when both surfactants were added to renovated water in concentrations higher than 1 CMC. The results obtained by Górna et al. [32] confirmed the potential of the MEUF process for the removal of biogenic substances, however, a cationic surfactant - ODA - gave better phosphate retention coefficients (9599%) than CPC (42–79% retention of phosphates). These findings supported the importance of all process parameters (involving the type of surfactant and its concentration, as well as membrane properties) on the efficiency of the MEUF process. Generally, there is limited literature available on fluoride removal using MEUF. Only one case study has been described in detail [20]. In the MEUF experiments, a cationic surfactant (CPC) at a varied concentration (ranging from 30 to 50 CMC) was used. The initial fluoride concentration amounted to 15 mg F-/L and cross-flow ultrafiltration was performed with a 10 kDa membrane at a transmembrane pressure of 276 kPa. It was shown that when the CPC concentration was higher than

36 mM, the removal efficiency was above 90%.It was also proved that in terms of flux decrease, the most optimal CPC concentration was 40.5 mM. The aim of this paper was to evaluate the usability of micellar enhanced ultrafiltration (MEUF) for fluoride removal from aqueous solutions with the use of two cationic surfactants (cetylpyridinium chloride - CPC and octadecylamine acetate - ODA). The effect of several operating parameters (membrane type, initial fluoride and surfactant concentration, mineral salt dose) on fluoride removal efficiency was established.

2. METHODS AND REAGENTS 2.1 Membranes In the course of the MEUF experiments, flat ultrafiltration membranes were used. Two types of Microdyn Nadir membranes were chosen - polyethersulfone (cut-off 4 kDa) and regenerated cellulose (cut-off 5 kDa). The characterization of the investigated membranes is given in Table 1. The chosen membranes were made of various polymers and varied in hydrophilicity and hydraulic permeability. Table 1. Membrane characterization Symbol Membrane material PES4 polyethersulfone CEL5 regenerated cellulose

Cut-off, kDa 4 5

Membrane character intermediate hydrophilic hydrophilic

The diameter of each membrane was equal to 76 mm, which corresponded to an effective surface area of 0.0045 m2. The low cut-off membranes were chosen because the most possible high separation efficiency was not only required for fluoride ions, but also in relation to surfactant monomers.The usage of high MWCO (molecular weight cut-off) membranes will result in retention coefficient deterioration. 2.2 Installation The MEUF experiments were conducted in a laboratory ultrafiltration Amicon 8400 cell. The overall volume of the UF cell amounted to 350 cm3 . The MEUF process was run at a transmembrane pressure of 0.2 MPa. The pressure was generated by a nitrogen gas cylinder. In order to maintain a constant concentration of the feed solution, the permeate was periodically recirculated by the recirculation pump. Prior to each cycle, the membrane was treated with distilled water at 0.2 MPa, until a constant permeate flux was established. All MEUF experiments were performed at room temperature. 2.3 Reagents 2.3.1 Fluoride and sodium chloride All the experimental series were performed with the use of solutions containing fluoride. Model solutions were made of sodium fluoride (molecular weight 41.99 Da). The F- ion content amounted to 10 and 100 mg F-/L. Experiments with fluoride solutions, supplemented with mineral salt (NaCl),were also performed. Sodium chloride concentration reached 0.5 and 1 g NaCl/L. All MEUF experiments were performed at a natural pH of the model solutions, which was in the range of 6-7. Sodium chloride was chosen as an example of monovalent salt, which is common in natural waters. This salt was used in the MEUF experiments and focused on the competitive removal of two ionic species. 2.3.2 Surfactants The MEUF tests were performed with the use of cationic surfactants: CPC (cetylpyridinium chloride) and ODA (octadecylamine acetate). The values of CMC, according to literature [30], are given in Table 2.

Table 2.Characterization of ODA and CPC surfactants [11, 30] Molecular CMC Molecular weight, Surfactant formula [mM] [mg/L] Da

ODA CPC 1

C20H43NO2 C21H38NCl

329 339

0.90 0.90

296.9 322.2

Molecular weight of micelle, Da 34,010

LogKow

6.991 3.461

ChemSrc (http://m.chemsrc.com/en/index.jsp)

The surfactants were obtained from Sigma Aldrich. During the tests with fluoride solutions deprived of sodium chloride, the CMC according to the literature value was used. In the case of experiments with an addition of NaCl, the CMC determined for a given fluoride-NaCl solution was applied. 2.4 Methodology The MEUF experiments were conducted for the model solutions in two stages. In the first stage, the treated solutions were composed of sodium fluoride and cationic surfactant (CPC or ODA). The surfactants were added in amounts that corresponded to 1 CMC, 2 CMC, or 3 CMC (the CMC value was taken from literature). In the second stage of experiments, the fluoride-surfactant mixtures were additionally supplemented with sodium chloride (NaCl) and the surfactant content was equal to 3 CMC. It should be underlined that due to a strong dependence of CMC on salt concentration, the CMC value was individually established for all the solutions used in the second stage of the MEUF experiments (Table 3). For comparison purposes, the CMC value was also determined for twocomponent solutions (fluoride-surfactant) with no salt addition. The surfactants were applied in moderate concentrations (1-3 CMC), because at higher CPC or ODA amounts there was a possibility of intensive and fast membrane fouling, which might result in a shortening of the membrane's life and a worsening of the membrane's hydraulic properties. At a high CMC, the risk of surfactant monomer leakage to the permeate could be also increased. The CMC value was determined based on the electrolytic conductivity measurements at various surfactant concentrations. The real CMC value was obtained at the intersection point of the linear relationships (conductivity versus CPC or ODA concentration) varying in slope. During the experiments, the electrolytic conductivity, fluoride concentration, total organic carbon content (TOC) and permeate volume flux were determined. The electrolytic conductivity was measured with the use of a Elmetron CC-411conductivity meter. The measurement accuracy was equal to ±0.2%. The fluoride concentration was analysedusing thecolorimetric method with SPADNS reagent. This reagent contains zirconium, which creates complexes with fluoride ions. The intensity of the sample color decreases when the fluoride content increases.A spectrophotometer HACH 2000 was used for fluoride analysis at a wavelength of 580 nm (program no. 190, method 8029). Measurement accuracy was equal to ±9%- this value was estimated on the basis of a relative error. The TOC content was monitored with the use of a HACH IL550 TOC-TN analyser. A method of catalytic high-temperature combustion was involved. The samples of permeate and feed solution (50 µ L) were acidified prior to analyses using 18.5% HCl acid. The permeate volume flux was calculated according to the following equation: (1) where: J- permeate volume flux (m3/m2day),V- volume of permeate (m3), t- time (day), A- surface area of the membrane (m2 ).

Generally, a steady water flux for a fresh/cleaned membrane was established after 2-3 h of transmembrane pressure operation (0.2 MPa).Following membrane conditioning with water, fluoride solution (with CPC or ODA and with or without NaCl) was passed through the membrane and the permeate flux was determined after 1h of the MEUF process. The fluoride retention coefficient R was calculated from the formula: (2) where: R - fluoride retention coefficient (%), Ci - initial fluoride content (mg F-/L), CP - fluoride content in the permeate (mg F-/L). The relative permeability, which is a measure of membrane sensibility to fouling, was calculated as the ratio J/J0, where: J - permeate volume flux (m3/m2 day) and J0 - distilled water flux (m3/m2 day). 3. RESULTS AND DISCUSSION 3.1 Influence of mineral salt on the value of CMC Before evaluation of the MEUF process, the values of CMC for CPC and ODA surfactants in multicomponent solutions were verified. The aim of these experiments was mainly focussed on the effect of sodium chloride on the CMC value. This value was determined by electrolytic conductivity measurements in relation to varied surfactant concentration (from 0.05 to 3 g/L). The obtained results are shown in Table 3. Table 3. CMC value versus fluoride and salt content Solution

CMC value for surfactant (according to literature [30]) 10 mg F-/L 10 mg F-/L + 0.5 g NaCl/L 10 mg F-/L+ 1 g NaCl/L 100 mg F-/L 100 mg F-/L+ 0.5 g NaCl/L 100 mg F-/L + 1 g NaCl/L 1 determined at room temperature (approx. 25 ºC)

CMC1 , g/L CPC 0.3222

ODA 0.2966

0.3189 0.1980 0.1023 0.2591 0.1547 0.1491

0.2626 0.2239 0.2045 0.2479 0.1950 0.1678

As was anticipated, the presence of mineral salt (NaCl) in the surfactant solution significantly influenced the CMC value of CPC and ODA. It was found that with increasing the concentration of sodium chloride, the CMC value decreased. This drop in CMC was more pronounced for the CPC surfactant – 1 g NaCl/L arrived at the CMC equal to 30-45% of the CMC in distilled water. It is interesting to note that even fluoride ions in low concentrations caused a 10% decrease of the CMC value for the ODA surfactant. Generally, in the case of ionic surfactants, the CMC is reduced by adding electrolytes [20]. Due to the presence of mineral salt, the electrostatic repulsion between the polar head groups is significantly diminished. Consequently,the micelles are formed easily at lower surfactant concentrations and the surfactants become more hydrophobic than in the solutions without electrolytes. In turn, with the increase of hydrophobic character of the surfactant, the aggregation number also increases [33]. The extent of binding of the counter ion increases with the charge of the counter ion, and decreases with the increase of its hydrated radius size [20]. Both counter ions (F- and Cl-) are characterized by a relatively small hydrated radius (136 and 181 nm [34], respectively), and therefore the screen effect of electrostatic repulsion occurs to a great extent. In brief - more micelles were formed after the addition of salt.

3.2 MEUF tests without salt addition 3.2.1 Influence of initial fluoride concentration, membrane material and surfactant type on fluoride separation efficiency The MEUF experiments were conducted for model solutions containing fluoride in amounts of 10 and 100 mg F-/L. In this series of experiments, the CPC and ODA concentration was equal to 1, 2 and 3 CMC. The final fluoride concentrations in the MEUF permeate and also the F- rejection coefficients versus the membrane type, surfactant concentration and initial fluoride concentration are presented in Figures 1 and 2. For a low fluoride content(10 mg F-/L),the best permeate, in view of F- ion content, was received when the CPC surfactant was present in the treated solutions at the highest applied concentration (3 CMC) and the PES4 membrane was applied. In this case, the fluoride concentration after the MEUF process reached 1 mg F-/L, thus meeting the permissible concentration level for drinking water (1.5 mg F-/L). Lowering the CPC concentration to 2 CMC still enabled a satisfactory fluoride concentration in the permeate (1.5 mg F-/L) for the PES4 membrane to be achieved. By using the CEL5 membrane and CPC surfactant, a slightly worse fluoride removal (in comparison to the PES4 membrane) was obtained – the retention coefficient amounted to 75-82% for 2 CMC and 3 CMC, respectively. The MEUF process with the ODA surfactant gave significantly worse results than MEUF with CPC – the fluoride retention coefficient varied from 57 to 74% (PES4 membrane) and from 21 to 73% (CEL5 membrane). Application of the ODA surfactant did not provide a F- ion concentration below 1.5 mg F/L, irrespective of the applied membrane and the CMC value. The observed final fluoride concentration after the MEUF process with the ODA surfactant was in the range of 2.6-4.3 mg F-/L (PES4 membrane) and 3.7-7.9 mg F-/L(CEL5 membrane). The poor fluoride rejection observed for octadecylamine acetate can be explained by ODA behaviour at high concentrations. It was confirmed that aqueous ODA solution containing 0.5-1 wt. % of surfactant exhibited pH in the range of 4.5-5.3 [35]. It can be anticipated that ODA concentration in the near-membrane surface layer is significantly increased (due to concentration polarization), thus strongly acidic conditions can be created. It was also proved that cationic surfactants are sensitive to pH variation and the CMC value may increase with the pH decrease [20, 36]. This phenomenon can be explained by surfactant protonation and the increase of the charge density at the micelle surface. This in turn increases the electrostatic repulsion between the charged heads and the micelle stability is changed.

Fig.1. Final fluoride concentration in the permeate and the F-retention coefficient versus membrane type and surfactant concentration: a) PES4 membrane, b) CEL5 membrane; Ci= 10 mg F-/L, ∆P = 0.2 MPa, error bars ±9%.

Fig.2. Final fluoride concentrationin the permeate and the F-retention versus membrane type and surfactant concentration: a) PES4 membrane, b) CEL5 membrane; Ci= 100 mg F-/L, ∆P = 0.2 MPa, error bars ±9%. In the case of the MEUF process conducted with solutions containing 100 mg F-/L, the obtained results were totally unsatisfactory for both membranes and both surfactants, irrespective of the CMC value. When the PES4 membrane was used,the final fluoride concentration in the permeate amounted to 49 – 65 mg F-/L (for the ODA surfactant) and to 57-74 mg F-/L during the tests with CPC. In this case, the corresponding fluoride rejection coefficients were rather low – 35-51% and 26-43% for the ODA and CPC surfactant, respectively.The MEUF process with the CEL5 membrane arrived at an even worse fluoride rejection than for the PES4 membrane. Application of the CPC surfactant allowed the F- content in the permeate to be decreased to 65-96 mg F-/L (the F-retention coefficient was equal to 4-35%), whereas the ODA surfactant enabled fluoride removal of 39-72% (the final F- ion concentration in the permeate was equal to 61 to 73 mg F-/L). Considering the MEUF results given in Figures 1-2, it can be concluded that the polyethersulfone membrane (PES4) demonstrated better separation properties towards fluoride ions than the cellulose membrane (CEL5). This finding can be attributed to a lower cut-off of the PES4 membrane in comparison with the CEL5 membrane. Moreover, the PES4 membrane is less hydrophilic than the CEL5 membrane, which means that aqueous solution with contaminants will pass through the CEL5 membrane more easily. It was also observed that the fluoride removal efficiency increased with an increasing CMC value. This univocal relationship was stated for both the applied surfactants. It is evident that a higher applied CMC value corresponds with more micelles that can be created. The more that micelles are present, the more fluoride ions that can be bound and removed from the treated solution. However, the rather poor fluoride retention coefficients obtained for the solutions containing elevated F-ion concentrations (100 mg F-/L) indicated that not enough micelles were formed and all the fluoride ions could not be bound. Possible, this was a reason that the enormous high CMC was applied (40-50 CMC) to remove more than 90% of fluoride by MEUF with CPC surfactant, as reported in literature [20]. It is also important to underline that a high surfactant concentration can create a gel layer at the membrane surface. This layer may operate as a “second membrane”, thus enhancing the fluoride removal efficiency, but on the other hand, it can significantly decrease the membrane volume flux. 3.2.2 Influence of initial fluoride concentration, membrane material and surfactant type on membrane permeability 3.2.2.1. Permeate flux In the course of the MEUF process,permeate flux was calculated according to equation (1). The obtained results are given in Figures 3 and 4.

Fig.3. Permeate flux versus membrane type and surfactant concentration: a) PES4 membrane, b) CEL5 membrane; Ci = 10 mg F-/L, ∆P = 0.2 MPa; standard deviation ± 0.02.

Fig .4. Permeate flux versus membrane type and surfactant concentration: a) PES4 membrane, b) CEL5 membrane; Ci = 100 mg F-/L, ∆P = 0.2 MPa; standard deviation ± 0.02. Prior to the tests with solutions containing fluoride and surfactant, the membrane permeability for distilled water was determined. The water volume flux amounted to 0.86 m3/m2 day for the CEL5 membrane and to 0.43 m3/m2 day for the PES4 membrane. The preferable permeabilityof the cellulose membrane when compared to the permeability of polyethersulfone membrane was attributed to the higher cut-off value and higher hydrophilicity of the CEL5 membrane in comparison to the PES4 membrane. In the course of MEUF process a general trend of membrane permeability decrease with increasing CMC value was observed. This relationship was more pronounced for ODA surfactant due to its lower solubility and relatively higher viscosity of ODA solution than CPC solution [27]. With the increase of surfactant feed concentration (> CMC) more micelles are formed, which are retained by membrane. The increased number of micelles in the near-membrane layer results in additional hydraulic resistance and the permeate flux declines. When the F- content amounted to 10 mg F-/L and the PES4 membrane was used, the water flux varied from 0.26 to 0.47 m3/m2day for the ODA surfactant and from 0.32 to 0.39 m3/m2day for the CPC surfactant. Application of the CEL5 membrane brought about a higher water flux, i.e. 0.38 - 0.84 m3/m2 day for the tests with the ODA surfactant and 0.77 - 0.86 m3 /m2 day during the experiments with the CPC surfactant. The elevated F-ion content in the treated solutions (100 mg F-/L) had a minor impact on membrane permeability. The permeate flux remained unchanged or even slightly decreased. In the course of MEUF with the PES4 membrane, the observed permeate flux amounted to 0.39- 0.44 m3/m2 day and 0.24-0.32 m3/m2day for the ODA and CPC surfactants, respectively. MEUF with the CEL5 membrane achieved a higher permeability (in comparison to the PES4 membrane). The permeate flux varied in the range of 0.52 - 0.61 m3/m2 day (for tests with the ODA surfactant) and 0.81 – 0.92 m3/m2 day (for tests with the CPC surfactant). It is interesting to note that in the latter case, the observed permeate

flux for the solution containing 100 mg F-/L was even slightly higher than for the solution containing 10 mg F-/L. The permeate flux drop for the contaminated solutions in comparison to the distilled water flux was caused by adsorption of surfactant monomers at the membrane surface or in the membrane pores. In the course of the MEUF process with the polyethersulfone membrane, the presence of the CPC surfactant caused a greater flux drop than for the ODA surfactant, irrespective of the fluoride concentration. The opposite phenomenon occurred in the case of the cellulose membrane, i.e. the ODA surfactant led to a higher permeate flux fall-off. However, it should be kept in mind that surfactants are able to modify the propertiesof a membrane surface by increasing its hydrophilicity [37]. It should also be mentioned that the real CMC value could be influenced by the presence of electrolytes. For a low fluoride concentration, the variations in the CMC values are minor for both surfactants, whereas the elevated F- ion content (100 mg F-/L) caused approximately a 17-20% decrease of the CMC value (Table 3). As a consequence, more micelles could be created under the presence of a high fluoride concentration than in the solutions with a low F- ion concentration. Thus, less fouling intensity caused by surfactant monomers could be expected in the course of MEUF with solutions containing 100 mg F-/L. In summary, it can be concluded that the permeate flux during MEUF of a fluoride-surfactant mixture is affected by many factors, such as interactions between a membrane and solution components, interactions between components of a solution between themselves, as well as the relation between the monomers and micelles that are present in a solution. 3.2.2.1. Relative permeability Membrane fouling can cause an increase in membrane hydraulic resistance and also a severe flux decline. The vulnerability of a membrane to fouling was estimated by calculating the relative permeability. The obtained results are given in Figures 5 and 6.

Fig.5. Relative permeability versus membrane type and surfactant concentration: a) membrane PES4, b) membrane CEL5; Ci = 10 mg F-/L, ∆P = 0.2 MPa.

Fig.6. Relative permeability versus membrane type and surfactant concentration: a) membrane PES4, b) membrane CEL5; Ci = 100 mg F-/L, ∆P = 0.2 MPa.

It was observed that the presence of surfactants resulted in a permeate flux fall-off in most cases. This phenomenon was especially pronounced for the ODA surfactant, as well as for solutions containing an elevated fluoride content (100 mg F-/L). As was already mentioned, with an increasing salt content (sodium chloride as well as sodium fluoride), the CMC value is decreased and more micelles are created. Thus, it seems that membrane fouling is mainly caused by the deposition of micelles or n-monomers on the surface of a membrane. This finding is also supported by a diminishing relative permeability with an increasing CMC value. At a low initial fluoride concentration (10 mg F-/L), the relative permeability varied from 0.60 to 1.09 (PES4 membrane) and from 0.44 to 1.0 (CEL5 membrane). The greatest drop in the relative permeability was noticed during the tests with the CEL5 membrane for the ODA surfactant at a concentration of 3 CMC (Fig. 5b). The highest relative permeability occurred for the PES4 membrane when the treated solution contained ODA at a concentration of 1 CMC. At an elevated fluoride concentration (100 mg F-/L), the relative permeability varied from 0.56 to 1.02 (PES4 membrane) and from 0.6 to 1.10 (CEL5 membrane). The greatest decrease in the relative permeability value occurred for the tests with the PES4 membrane for the CPC surfactant at a concentration of 3CMC. The highest relative permeability was observed for the CEL5 membrane when 2 CMC tests of the ODA surfactant were conducted (Fig. 6b). When taking into account the obtained results, it can be concluded that there is no univocal relationship between the membrane type and fouling intensity. Membrane vulnerability to fouling depends on the surfactant type, its concentration, as well as the salt content in the treated solution. In a few cases it was observed that the relative permeability was higher than 1 (i.e. the permeate flux of the treated solution was higher than the water flux). This phenomenon could be caused by hydrophilization of the membrane surface by the surfactant, which allowed for the transportation of the aqueous solution through the membrane more easily. 3.3 MEUF tests with the addition of mineral salt 3.3.1 Influence of sodium chloride on the fluoride retention coefficient Small quantities of sodium chloride (0.5 and 1.0 g NaCl/L) were added to the fluoride-surfactant mixtures to investigate the effect of this salt on the removal of F- ions from aqueous solutions. The obtained final concentrations and retention coefficients of the F- ions are shown in Figures 7 and 8.

Fig.7. Final fluoride concentration in the permeate and the F- retention coefficient versus the membrane and surfactant type, and NaCl concentration: a) membrane PES4, b) membrane CEL5; Ci = 10 mg F-/L, ∆P = 0.2 MPa; error bars ±9%.

Fig.8. Final fluoride concentration in the permeate and the F- retention coefficient versus the membrane and surfactant type, and NaCl concentration: a) membrane PES4, b) membrane CEL5; Ci = 100 mg F-/L, ∆P = 0.2 MPa; error bars ±9%. When taking into consideration the results of the F- rejection given in Figs. 1 and 2, and Figs. 7 and 8, it is quite clear that the presence of NaCl brought about a dramatic decrease of MEUF separation efficiency in fluoride ions. This trend was observed for all the experimental series, irrespective of membrane and surfactant type. Moreover, the decrease in fluoride removal was intensified with an increasing NaCl content in the case of a low initial fluoride content (10 mg F-/L) (Fig. 7), whereas the opposite tendency was monitored for solutions containing a high amount of F- ions (100 mg F-/L) (Fig. 8). When comparing the fluoride separation by both membranes, the PES4 membrane exhibited a better retention coefficient than the CEL5 membrane. These results were quite similar to the results obtained for the experiments with a lack of NaCl in the treated solutions. However, the retention coefficients reached under the presence of sodium chloride were too low to give appropriate permeate quality in view of fluoride content ( < 1.5 mg F-/L). For a low fluoride content (10 mg F-/L), the best permeate, in view of the F- ion content, was received when the CPC surfactant was present in the treated solutions and a low salt content was applied (0.5 g NaCl/L) – the final fluoride concentration amounted to 5.45 mg F-/L (PES4 membrane) and to 6.25 mg F-/L (CEL5 membrane). The increase of NaCl concentration to 1.0 g NaCl/L resulted in a significant deterioration of F- ion separation efficiency – the final fluoride content was equal to 6.95 and 7.9 mg F-/L for the PES4 and CEL5 membranes, respectively. Application of the ODA surfactant gave poorer results than those obtained for the CPC surfactant. The observed F- ion retention coefficients were in the range of 32-34% (PES4 membrane) and 20.5-23% (CEL5 membrane), which corresponded with the final fluoride content in the range of 6.55-6.8 and 7.7-7.95 mg F-/L for the PES4 and CEL5 membranes, respectively. MEUF experiments with solutions containing high fluoride concentrations (100 mg F-/L) resulted in a subsequent deterioration of F- ion separation efficiency. Generally, the retention coefficients were in the range of 18-28.5% (PES4 membrane) and 5-20% (CEL5 membrane). By applying the ODA surfactant, insignificantly better results were obtained than with the CPC surfactant, however the lowest obtained concentration of fluoride (71.5 mg F-/L) makes the MEUF process rather unsuitable for the removal of high fluoride amounts from NaCl-surfactant solutions. It is interesting to note that the increase of NaCl dosage (to 1g NaCl/L) slightly improved the MEUF separation of fluoride for both tested membranes, as well as for both used surfactants. The observed unfavourable effect of sodium chloride on the fluoride removal efficiency using the MEUF process can be attributed to the competition between F- and Cl- ions. As a result, consumption of the available binding sites of the micelles by co-existing anions (Cl-) might have occurred. This supposition is supported by Chen et al. [38],who proved that the affinity of F- ions for the CPC micelles is lower than the affinity of Cl- anions. Iqbal et al. [27] confirmed a minor drop (4-8%) in arsenate removal under the presence of small quantities of nitrate and phosphate, irrespective of the type of cationic surfactant used (CPC, ODA or CTAP). In turn, Gzara at el. [28] discovered that

chromate rejection in the MEUF process with the CPC surfactant remained constant (approx. 90%), even in the presence of 100 mM of NaCl. However, they applied the theoretical CMC value for the CPC surfactant, and thus the decreased CMC could be expected in the NaCl solution (more micelles were formed in the NaCl solution than in a solution free of NaCl). It should also be mentioned that after the addition of salt, the content of monovalent counterion increases and this leads to compression of the double layer that surrounds the micelles. As a result, at any given distance from the micelle surface, the absolute magnitude of electrical potential is decreased. The lowered electrical potential results in a reduction of driving force for the adsorption (i.e. electrostatic attraction) of anions at the micelle surface of a cationic surfactant [39]. 3.3.2 Influence of sodium chloride on membrane permeability 3.3.2.1 Permeate flux In the course of the MEUF process with multi-component solutions (fluoride-surfactant-sodium chloride mixtures), the permeate flux was determined. The calculated values of the permeate flux in relation to the NaCl content and initial F- concentration are shown in Figures 9 and 10.

Fig.9. Permeate flux versus membrane and surfactant type, and NaCl content: a) PES4 membrane, b) CEL5 membrane; Ci = 10 mg F-/l, ∆P = 0.2 MPa; standard deviation ± 0.02.

Fig.10. Permeate flux versus membrane and surfactant type, and NaCl content: a) PES4 membrane, b) CEL5 membrane; Ci = 100 mg F-/L, ∆P = 0.2 MPa, standard deviation ± 0.02. As in the MEUF tests performed for the fluoride solutions free of NaCl, there was a deterioration of the permeate flux for contaminated solutions when compared to the water flux. However, the drop of membrane permeability for the three-component mixtures and the ODA surfactant was by

approximately 20-40% greater when compared to the membrane flux for the two-component solutions (fluoride-surfactant). This severe decline of flux for the solutions containing ODA was associated with a lower water solubility and the relatively higher viscosity of the ODA solutions when compared to the CPC solutions [27]. When analysing the permeate flux in terms of membrane type, the PES4 membrane exhibited poorer permeability than the CEL5 membrane, especially for the solutions containing 10 mg F-/L (Fig. 9). This finding was basically in accordance with the previous experiments(no NaCl addition). For the ODA and CPC surfactants, the permeate flux amounted to 0.24 m3/m2day and 0.20-0.22 m3/m2 day, respectively (PES4 membrane), whereas permeability of the CEL5 membrane amounted to 0.46-0.49 m3/m2 day (ODA surfactant) and 0.79-0.96 m3 /m2day (CPC surfactant). A distinct tendency of a decrease in flux with an increased NaCl concentration was seen for all the experimental series. At an elevated fluoride content (100 mg F-/L), the observed relationships of permeate flux versus surfactant and membrane type also reflected earlier dependences (Fig. 4). The permeate flux of the PES4 membrane amounted to 0.31 and 0.28 m3/m2 day when the ODA surfactant was applied and the salt content was equal to 0.5 and 1 g NaCl/L, respectively. The usage of the CPC surfactant gave the PES4 membrane permeability in the range of 0.23 and 0.21 m3/m2 day for 0.5 and 1 g NaCl/L, respectively. However, the flux decline for MEUF performed by the ODA surfactant and the CEL5 membrane was unexpectedly large - more than a two-fold flux decrease in comparison to the water flux was stated (Fig. 10b). This explicit decrease of membrane permeability under the presence of sodium chloride could probably be attributed to the shielding effect of the electrostatic repulsion between micelles, as well as between the membrane surface charge and the solution components. Moreover, the added NaCl makes the surfactants effectively more hydrophobic [33], thus increasing the fouling potential in the MEUF process. However, the CPC surfactant exhibited the lowest fouling potential in MEUF with the CEL5 membrane. Generally, the decrease of membrane permeability is caused by surfactant deposition at the membrane surface and/or in the membrane pores. As in the results obtained in the previous tests, it was observed that the PES4 membrane was fouled to a greater extent by the CPC surfactant, whereas the CEL5 membrane was blocked by ODA. It seems that the observed differentiations in membrane permeability could also be attributed to the hydrophilicity/hydrophobicity of the surfactants – the ODA surfactant is more hydrophobic than the CPC surfactant (Table 1), and can therefore be deposited in the CEL5 membrane (with greater cut-off than the PES4 membrane) more easily. 3.3.2.1 Relative permeability The membrane sensibility to fouling under the presence of sodium chloride was evaluated by calculating the relative permeability. Figures 11 and 12 reflect the permeability of membranes CEL5 and PES4.

Fig.11. Relative permeability versus membrane and surfactant type, and NaCl content: a) PES4 membrane, b) CEL5 membrane; Ci = 10 mg F-/L, ∆P = 0.2 MPa.

Fig.12. Relative permeability flux versus membrane and surfactant type, and NaCl content: a) PES4 membrane, b) CEL5 membrane; Ci = 100 mg F-/L, ∆P = 0.2 MPa. In principle, the data given in Figures 11 and 12 exactly reflects the membrane permeability discussed in the previous chapter(3.3.2.1). In almost all cases, the relative permeability under the presence of NaCl was lower than the relative permeability calculated for the solutions free of sodium chloride. The relative permeability was insignificantly higher for the CEL5 membrane and low fluoride concentration (10 mg F-/L) than for the two-component solution (Fig.5b and Fig.11b). However, salt addition in the amount of 1 g/L for the CPC surfactant resulted in a considerable drop in the relative permeability (from 1.12 to 0.77). Moreover, this adverse effect of the increased NaCl content on the relative permeability was observed for all the experimental series. As was already mentioned, salt (e.g. sodium chloride) facilitates aggregation of ionic surfactants due to the electrostatic shielding effect and denser gel layer that can be formed at the membrane surface. For a low fluoride content (10 mg F-/L), the relative permeability for the PES4 membrane amounted to 0.56 (ODA surfactant) and 0.47-0.50 (CPC surfactant). The CEL5 membrane exhibited relative permeability in the range of 0.53-56 (ODA surfactant) and 0.77-1.12 (CPC surfactant). After increasing the fluoride content to 100 mg F-/dm3 , the relative permeability for the PES4 membrane varied from 0.49 to 0.72. The lowest value was noticed during the tests with the CPC surfactant and the solution containing 1 g NaCl/L, whereas the highest relative permeability occurred when the treated solution contained ODA and 0.5 g NaCl/L. It should be underlined that an extremely low relative permeability (0.35-0.36) was observed for the cellulose membrane in the course of the MEUF process with the ODA surfactant, whereas application of the CPC surfactant resulted in a high value of relative permeability (0.91 – 0.92). These differences in relative permeability might be attributed to the diverse hydrophobicity of the surfactants. 3.3.3 TOC content One serious problem in the MEUF process is the leakage of the surfactant to the permeate. Surfactant monomers are too small to be retained by UF membranes and they cause undesirable pollution of the MEUF effluent. The extent of this phenomenon for the MEUF process with threecomponent solutions was verified by analysing the TOC content of the permeate. The permeate TOC values in relation to fluoride and NaCl concentration for the CPC and ODA surfactants are given in Table 4.

Table 4. TOC concentration of MEUF permeates in relation to various composition of feed solution and membrane type TOC in permeate, mg C/L Initial TC, Composition of the feed PES4 membrane CEL5 membrane mg C/L solution mg/L %* mg/L %* 10 mg F-/L + 0.5 g NaCl/L 342.2 58.89 17.2 60.68 17.7 + 3 CMC CPC 10 mg F /L + 0.5 g NaCl/L 314.4 63.67 20.3 64.81 20.6 + 3 CMC ODA 10 mg F /L + 1 g NaCl/L + 342.2 39.00 11.4 40.00 11.7 3 CMC CPC 10 mg F /L + 1 g NaCl/L + 314.4 55.61 17.7 58.56 18.6 3 CMC ODA 100 mg F /L + 0.5 g NaCl/L 342.2 49.69 14.5 54.61 16.0 + 3 CMC CPC 100 mg F /L + 0.5 g NaCl/L 314.4 51.03 16.2 53.02 16.9 + 3 CMC ODA 100 mg F /L + 1 g NaCl/L 342.2 32.77 9.6 37.83 11.1 + 3 CMC CPC 100 mg F /L + 1 g NaCl/L 314.4 51.30 16.3 53.04 16.9 + 3 CMC ODA * % of initial surfactant amount transported to permeate TOC for 1 CMC CPC solution: 137.4 mg C/L TOC for 1 CMC ODA solution: 116.5 mg C/L It was observed that the TOC content in the permeate for all the experimental series varied in the range of approximately 33-65 mg C/L. The initial TOC concentration in the feed solutions amounted to 342.2 mg C/L (for CPC) and 314.4 mg C/L (for ODA). Thus, it can be estimated that about 10-16% of the initial TOC occurred in the permeates. Interestingly, the quality of received permeates in view of the TOC value are surprisingly similar, indicating that both membranes allowed surfactant monomers to pass into the permeate side. This finding can be supported by the observed TOC concentration in permeates – it was much lower than TOC determined for solutions containing ODA and CPC in the amount of 1 CMC (Table 4). Insignificantly higher TOC concentrations in the MEUF permeates were observed for the ODA surfactant when compared to the CPC surfactant, which can be attributed to the higher real CMC value for the ODA in the NaCl solutions than for the CPC (Table 3). It was also found that with an increasing salt content in the feed, the permeate TOC content decreased (especially for the CPC surfactant). This effect can be explained by the facilitating aggregation of the surfactants under the presence of electrolytes. According to literature [28], the electrolyte dosage to the MEUF feed can be considered as a method of diminishing the surfactant content in the permeate. Gzara et al. [28] revealed that increasing the NaCl content from 1 to 500 mM resulted in a decrease of the CPC concentration in the permeate from 1 to 0.15 Mm. However, it should be remembered that the real CMC values for solutions containing salt were applied in the present study, and therefore the salt effect was not so spectacular. With the aim of improving the permeate quality and reducing the surfactant content, some posttreatment methods should be involved. Iqbal et al. [27] proposed the post treatment of the permeate using the PAC process. Electrodialysis with mono-ion selective ion-exchange membranes seems to also be suitable to separate surfactants in low concentrations from mineral monovalent salts. Another possible way to solve the problem of permeate contamination by surfactant monomers is by choosinga surfactant with a low CMC.

4. CONCLUSIONS • Micellar enhanced ultrafiltration (MEUF)enabled fluoride removal in drinking water to a permissible level (< 1.5 mg F-/L) on the condition that the feed fluoride concentration was below 10 mg F-/L and the cationic surfactant - cetylpyridinium chloride (CPC) was applied at a concentration of 3 CMC. The increased fluoride concentration (100 mg F-/L) gave less promising results, irrespective of the concentration and characteristics of the surfactant. It can be concluded that not enough micelles were formed in order to bind all the fluoride ions. • The fluoride retention coefficient increased with an increasing surfactant concentration. A higher concentration of applied surfactant corresponds with more micelles being created and more fluoride ions being bound and removed from the treated solution. The formation of a surfactant gel layer at the membrane surface can enhance fluoride removal efficiency. • A significant deterioration of membrane permeability in comparison to the water flux was observed in the course of the MEUF process with cetylpyridinium chloride (CPC) and octadecylamine acetate (ODA). The permeate flux during MEUF of a fluoride-surfactant mixture is affected by many factors, such as interactions between membrane and solution components, interactions between components of a solution between themselves, as well as the relation between monomers and micelles that are present in a solution. • The presence of a mineral salt (NaCl) in a surfactant solution significantly diminished the CMC value of CPC and ODA. This drop in CMC was more pronounced for the CPC than for the ODA surfactant.The decrease of the CMC value under the presence of sodium chloride was caused by diminishing the electrostatic repulsion between the polar head groups. • The MEUF process performed under the presence of sodium chloride in the treated solutions resulted in a significant deterioration of F- ion separation efficiency. This unfavourable effect of sodium chloride on fluoride removal using the MEUF process can be attributed to the competition between F- and Cl- ions. • Secondary contamination of the permeate in the course of the MEUF process was observed. This undesirable worsening of permeate quality was caused by the passing of surfactant monomers through UF membranes (leakage of the surfactant to the permeate). • It is recommended to use a surfactant characterised by a low CMC. This allows for a reduction of surfactant leakage to the permeate.

Acknowledgements The financial support of the Faculty of Environmental Engineering (Wroclaw University of Science and Technology, grant no. 0402/0080/16) is greatly appreciated. List of symbols: A - surface area of membrane (m2), Ci - initial fluoride content (mg F-/L), Cp - fluoride content in permeate (mg F-/L), CEL5 - regenerated cellulose membrane, cut-off 5 kDa, J - permeate flux (m3/m2 day), J0 - distilled water flux (m3/m2day), PES4 - polyethersulfone membrane, cut-off 4 kDa, R - retention coefficient (%), T - time (day), V - volume of permeate (m3). Highlights • Micellar enhanced ultrafiltration was applied for fluoride removal • MEUF enabled fluoride concentration in the permeate to be below the drinking water limit

• A meaningful flux decline was caused by surfactant deposition at the membrane surface • Monovalent salt deteriorated fluoride removal due to competition between Cl- and F- ions • The critical micelle concentration of cationic surfactants was reduced by adding mineral salt (NaCl)

References 1. I. Bejaoui, A. Mnif, B. Hamrouni, Performance of reverse osmosis and nanofiltration in the removal of fluoride from model water and metal packaging industrial effluent, Sep. Sci. Technol. 49 (2014) 1135–1145. 2. S. Lahnid, M. Tahaikt, K. Elaroui, I. Idrissi, M. Hafsi, I. Laaziz, Z. Amor, F. Tiyal, A. Elmidaoui, Economic evaluation of fluoride removal by electrodialysis, Desalination 230 (2008) 213–219. 3. A. Boubakri, N. Helali, M. Tlili, M.B. Mohamed, Fluoride removal from diluted solutions by Donnan dialysis using full factorial design, Korean J. Chem. Eng. 31(3) (2014) 461–466. 4. J. Zhu, X. Lin, P. Wu, Q. Zhou, X. Luo, Fluoride removal from aqueous solution by Al(III)-Zr(IV) binary oxide adsorbent, Appl. Surf. Sci. 357 (2015) 91–100. 5. J. Shen, A.Schäfer, Removal of fluoride and uranium by nanofiltration and reverse osmosis: A review, Chemosphere 117 (2014) 679–671. 6. I. Ali, Z.A. Alothman, M.M. Sanagi, Green Synthesis of iron nano-impregnated adsorbent for fast removal of fluoride from water, J. Mol. Liq.211 (2015) 457–465. 7. S. Chatterjee, S. De, Adsorptive removal of fluoride by activated alumina doped cellulose acetate phthalate (CAP) mixed matrix membrane, Sep. Purif. Technol.,125 (2014) 223–238. 8. R.O. Dunn, J.F. Camehorn, S.D. Christian, Use of micellar-enhanced ultrafiltration to remove dissolved organics from aqueous streams, Sep. Sci. Technol. 20 (1985) 257–284. 9. R. Ramona, M. Geanta, O. Ruiz, I. Escudero, Micellar-enhanced ultrafiltration for the recovery of lacticacid and citricacid from beet molasses with sodium dodecylsulphate, J. Membr. Sci. 430 (2013) 11–23. 10. E. Abbasi-Garravand, C.N. Mulligan, Using micellar enhanced ultrafiltration and reduction techniques for removal of Cr(VI) and Cr(III) from water, Sep. Purif. Technol. 132 (2014) 502–512. 11. J.L. Acero, F.J. Benitez, F.J. Real, F. Teva, Removal of emerging contaminants from secondary effluents by micellar-enhanced ultrafiltration, Sep. Purif. Technol. 181 (2017) 121–131. 12. P. Yenphan, A. Chanachai, R. Jiraratananon Experimental study on micellar-enhanced ultrafiltration (MEUF) of aqueous solution and wastewater containing lead ion with mixed surfactants, Desalination 253 (2010) 30–37. 13. J. Huang, L. Peng, G. Zeng, X. Li, Y. Zhao, L. Liu, F. Li, Q. Chai, Evaluation of micellar enhanced ultrafiltration for removing methylene blue and cadmium ion simultaneously with mixed surfactants, Sep. Purif. Technol. 125 (2014) 83–89. 14. T. Mehling, A. Zewuhn, T.Ingram,I. Smirnova, Recovery of sugars from aqueous solution by micellar enhanced ultrafiltration, Sep. Purif. Technol. 96 (2012) 132–138. 15. B. Tanhaei, M. PourafshariChenar, N. Saghatoleslami, M. Hesampour, T. Laakso, M. Kallioinen, M. Sillanpää, M. Mänttäri, Simultaneous removal of aniline and nickel from water by micellarenhanced ultrafiltration with different molecular weight cut-off membranes, Sep. Sci. Technol., 124 (2014) 26–35. 16. U. Danis, C. Aydiner, Investigation of process performance and fouling mechanisms in micellar-enhanced ultrafiltration of nickel-contaminated waters, J. Hazard. Mater.162 (2009) 577–587. 17. D. Doulia, I. Xiarchos, Ultrafiltration of micellar solutions of nonionic surfactants with or without alachlor pesticide, J. Membr. Sci. 296 (2007) 58–64. 18. R. Bade, S.H. Lee, A review of studies on micellar enhanced ultrafiltrationfor heavy metals removal from wastewater, Journal of Water Sustainability, 1 (2011) 85–102. 19.J. Landaburu-Aguirre, V. García, E. Pongracz, R.L. Keiski, The removal of zinc from synthetic wastewaters by micellar-enhanced ultrafiltration: statistical design of experiments, Desalination, 240 (2009) 262–269.

20. S. De, S. Mondal, Micellar Enhanced Ultrafiltration: Fundamentals & Applications, CRC Press, Boca Raton, 2012. 21. B. Tanhaei, M.PourafshariChenar, N. Saghatoleslami, M. Hesampour, M. Kallioinen, M. Sillanpää, M. Mänttäri, Removal of nickel ions from aqueous solution by micellar-enhanced ultrafiltration using mixed anionic–non-ionic surfactants, Sep. Purif. Technol.138 (2014)169–176. 22. S.H. Lee, S. Shrestha, Application of micellar enhanced ultrafiltration (MEUF) process for zinc (II) removal in synthetic wastewater: Kinetics and two-parameter isotherm models, Int. Biodeterior. Biodegradation 95 (2014) 241–250. 23. X. Zhang, L. Niu, F. Li, X. Zhao, H. Hu, Enhanced rejection of cations by low-level cationic surfactant during ultrafiltration of low-level radioactive wastewater, Sep. Purif. Technol. 2017 (175) 314–320. 24. J.-H. Huang, L.-J. Shi., G.-M. Zeng, X.Li, S.-B. He, F. Li, Y.-L. Xiong, S.-H. Guo, D.-M. Zhang, G.-X. Xie, Effects of feed concentration and transmembrane pressure on membrane fouling in Cd2+ removal by micellar-enhanced ultrafiltration, Desalination 294 (2012) 67–73. 25. Y.-Y. Fang, G.-M. Zeng, J.-H.Huang, J.-X. Liu, X.-M. Xu, K.Xu, Y.-H. Qu, Micellar-enhanced ultrafiltration of cadmium ions with anionic-nonionic surfactants, J. Membr. Sci.320 (2008)514–519. 26. J.-H. Huang,S.-H. Guo, G.-M. Zeng, Y.-L.Xiong, D.-M. Zhang.X.-J.Tang, G.-X. Xie, Prediction of fouling resistance and permeate flux in cross-flow micellar-enhanced ultrafiltration (MEUF), Colloids Surf. A: Physicochem. and Engineering Asp. 401 (2012) 81–89. 27.J. Igbal, H-J. Kim,J-S.Yang,K. Baek, J-W.Yang,Removal of arsenic from groundwater by micellarenhanced ultrafiltration (MEUF), Chemosphere 66(5) (2007) 970-976. 28.L. Gzara, M. Dhahbi, Removal of chromate anions by micellar-enhanced ultrafiltrationusing cationic surfactants, Desalination 137 (2001) 241–250. 29.M.K. Purkait., S.Dasgupta S.De,Micellar enhanced ultrafiltration of eosin dye using hexadecyl pyridinium chloride, J. Hazard. Mater. 136(3) (2006) 972–977. 30.R. Camarillo, I. Asencio, J. Rincon, Micellar Enhanced ultrafiltration for phosphorus removal in domestic wastewater, Desalin. and Water Treat. 6 (2009) 211–216. 31.K. Baek, B.-K. Kim,J.-W. Yang, Application of micellar-enhanced ultrafiltration for nutrients removal,Desalination, 156 (2003) 137–144. 32. J. Górna, K. Majewska-Nowak, Surfactant enhanced ultrafiltration for the removal of inorganic phosphorous compounds from aqueous solutions, Environment Protection Engineering 39(3) (2013) 197–206. 33. N. Ataci,A. Sarac, Determination of Critical Micel Concentration of PEG-10 Tallow Propane Amine: Effects of Salt and pH, Am. J. Analyt. Chem. 5 (2014) 22–27. 34. M. Tahaikt, R. El Habbani, A. AitHaddou, I. Achary, Z. Amor, M. Taky, A. Alami, A. Boughriba, M. Hafsi, A. Elmidaoui, Fluoride removal from groundwater by nanofiltration.Desalination 212(1-3) 2007) 46-53. 35. A.D. Mercer, Corrosion Inhibitors: Principles and Practice, in: Corrosion, vol.2, Corrosion Control, edited by L.L. Shreir; Newnes-Butterworths, London-Boston 1976 36. NeseAtaci, AyferSarac, Determination of critical micel concentration of PEG-10 tallow propane amine; effects of salt and pH, American Journal of Analytical Chemistry, 2014, 5, 22-27 37.S.M. Mousavi, S. Saljoughi, Z. Ghasemipour, S.A. Hosseini, Preparation and characterization of modified polysulfone membranes with high hydrophilic property using variation in coagulation bath temperature and addition of surfactant, Polym. Eng. Sci. 52(10) (2012) 2196-2205. 38. M. Chen, C.T. Jafert, Anion exchange on cationic surfactant micelles, and a speciation model for estimating anion removal on micelles during ultrafiltration of water, Langmuir, Article ASAP, DOI: 10.1021/acs.langmuir.7b01270 39.S.B. Kamble, K.V. Marathe, Micellar enhanced ultrafiltrration of chromate [Cr(VI)] ion from aqueous streams by using cationic surfactant, Indian J. Chem. 12 (2005) (12) 393–400.