**Textile Wastewater Purification Using an Elaborated Biosorbent Hybrid Material (***Lu*ff*a–Cylindrica***–Zinc Oxide) Assisted by Alternating Current**

#### **Amina Othmani 1,2, Aida Kesraoui 1, Roberto Boada 3, Mongi Se**ff**en 1,\* and Manuel Valiente 3,\***


Received: 31 May 2019; Accepted: 22 June 2019; Published: 27 June 2019

**Abstract:** This paper aims to synthesize hybrid materials with high pollutant-uptake capacity and low costbased based on *Lu*ff*a cylindrica* (*L.C*) and different percentage of Zn2<sup>+</sup> in the presence and absence of alternating current (AC). Physico-chemical, morphological and structural characterizations of the hybrid materials were performed by Boehm method, point zero charge (pHpzc), infrared characterizations (IR), scanning electron microscopy (SEM), energy–dispersive spectroscopyand and X-ray photoelectron spectroscopy. The efficiency of the designed hybrid materials was optimized based on their performance in water depollution. Methylene blue (MB) and industrial textile wastewater were the investigated pollutants models. IR characterizations confirmed the fixation of Zn2<sup>+</sup> onto the *L.C* by the creation of Zn-OH, Zn-O and Zn-O-C bonds. Boehm titration showed that the fixation of Zn2<sup>+</sup> onto *L.C* is accompanied by an increase of the basic functions of its surface and subsequently an increase in the pHpzc. SEM results confirmed the fixation of Zn2<sup>+</sup> onto the *L.C* coupling AC with biosorption showed an increase in the adsorbed amount of MB and speed when adding the 4% of Zn2<sup>+</sup> compared to the pure *L.C* the Qm shifted from 3.22 to 9.84 and 8.81 mg/g, respectively, for hybrid materials synthesized under AC, in absence of AC and pure *L.C.*

**Keywords:** alternating current; coupling; hybrid material; biosorption; wastewater reuse

#### **1. Introduction**

Textile dying wastewaters are classified among the most highly toxic effluents [1,2]. Numerous processes have been suggested for treatment and purification such as biological treatment [3], coagulation-flocculation [4], adsorption [5], ultrafiltration [6], electrocoagulation [7], reverse osmosis [8] and anodic oxidation [9]. The current research was devoted to water remediation through the valorization of the abundant renewable resource of cellulosic fibers [10]. Brown algae [11], *Bacillus maceran* [12], *Posidonia oceanica* [13], corn stigmas [14], *Agave Americana* [15], *Lu*ff*a cylindrica* [16], fly ash and red mud [17], raw date pits [18] and *Phragmites australis* [19] are widely used as a natural and cheap biosorbent for a pollutant removal. However, achieving a quick adsorption kinetic with a high possibility of water reuse after filtration absolutely depends on the biosorbent efficiency.

Hybrid materials are praised for not only intermediate properties between mineral and organic matter but also for new interesting behaviors that allow them to be used in several fields of applications. Such domains include catalytic applications, medical and pharmaceutical applications, optoelectronics, the environment, and biomaterials [20,21]. Hybrid materials can be prepared by several methods such as chemical vapor deposition (CVD) [22], physical vapor deposition (PVD) [23], laser ablation and

sputtering [24]. The synthesis of hybrid materials based on cellulose and zinc oxide is the objective of recent studies. Perelshtein et al. (2009) developed a composite based on cotton fibers and zinc oxide. They synthesized the ZnO particles and then deposited them on the cotton surface by ultrasonic irradiation (sonochemical method) [25]. Weili et al. (2010) have succeeded in synthesizing a hybrid matrix based on wurtzite-cellulose ZnO bacteria by the thermal decomposition method. This new hybrid material obtained plays a very important role in improving the photo catalytic activity of anionic dyes such as, orange methyl [26].

However, the techniques used for the synthesis of hybrid materials are expensive. Other researchers have developed less expensive techniques such as the electrochemical method [27], spray pyrolysis [28] and sol-gel process [29]. Recently, Kesraoui et al. (2018) reported that the precipitation method can be a simple and cheap method used for the synthesis of hybrid materials. They have successfully synthesized hybrid material base on *L.C* and metal oxides (ZnO, Al2O3). These last showed a good ability for dyes removals compared to the pure *Lu*ff*a cylidnrica* (*L.C*) [30]. According to literature, the synthesis of hybrid materials based on compounds containing cellulose and ZnO can present interesting properties allows it to be used in several applications.

However, the removal process often requires a lot of time to reach the balance. Therefore, coupling environmental security (there is no transformation of the initial molecule to toxic compounds as for the oxidation process), low cost, final efficiency, and quick process remains a very important stake.

Taking into account these considerations, a set of aims are proposed in this paper, the synthesis of a high performing hybrid material based on *Lu*ff*a cylinrdica* (*L.C*) and metal oxide with high pollutant-uptake capacity and low cost by an easy precipitation method in absence and presence of alternating current (AC). Since their good affinity for the removal of several pollutants and their interesting physical and chemical properties, *L.C* and ZnO were considered for the synthesis of the hybrid materials [31,32]. Two main pollutants were taken as models in this study; a cationic dye the methylene blue (MB) taken as a model of dye and an industrial textile wastewater. A detailed investigation of the physicochemical, morphological and structural characterizations was carried out using Boehm titration. Also, Drift method, scanning electron microscopy (SEM), Energy-dispersive spectroscopy (EDS), infrared spectroscopy (IR) and X-ray photoelectron spectroscopy (XPS) analysis were performed. The possible pathway of the Zn2<sup>+</sup> fixation into the *L.C* fibers and the mechanism of possible interaction of MB onto the lignocellulosic surface are suggested. The mathematic modeling was also supplied using the stochastic model of Brouers-Sotolongo.

#### **2. Material and Methods**

#### *2.1. Materials*

Methylene blue (MB) (purity: 95%) from Fluka™ (manufacturer, Streinheim, Germany), with chemical formula C16H18N3SCl and molar mass 319.85 g/mol, was chosen as commercial dye model for this study.

Zinc nitrate (Zn(NO3)2.6H2O; Hexahydrate Purified), sodium chloride (NaCl), sodium bicarbonate (NaHCO3), sodium carbonate (Na2CO3) and sodium hydroxide (NaOH) (purity 99%) were obtained from LOBA CHEMIE (Wodehouse road, Mumbai, India).

The biosorbent used for the synthesis of the hybrid material was fruits of Tunisian *Lu*ff*a cylindrica* fibers. This biosorbent was purchased at the local market Sousse. The chemical composition of these fibers was determined by Kesraoui et al. (2016). These fibers are composed of 54% of cellulose, 11% of lignin, 5% of pectin, 7% of fats and waxes and 23% hemicelluloses [33].

#### *2.2. Methods*

#### 2.2.1. Biosorbent and Adsorbate Preparation

Tunisian *Lu*ff*a cylidnrica* (*L.C*) was chosen as a natural biosorbent in this paper. The preparation of this biosorbent consisted of cutting the fibers finely, washing them several times to remove all impurities and drying them at 70 ◦C until the material was completely dried. As for the adsorbate preparation, it consisted of dissolving 10 mg of MB in 1 L of distilled water to obtain the desired concentration (10 mg/L).

#### 2.2.2. Preparation of Hybrid Materials *L.C*+ (1%, 2%, and 4% Zn2+) in Presence and Absence of AC

The Zn2<sup>+</sup> precursors of solution were obtained from (Zn(NO3)2·6H2O (Hexahydrate Purified) Sigma-Altrich (99%) (Wodehouse road, Mumbai, India).

The preparation of the hybrid material with different percentages of Zn2<sup>+</sup> (1%, 2%, and 4%) consists inmixing 5 g of *L.C* (size 250 μm) with 0.05, 0.1 and 0.2 g of Zn(NO3)2·6H2O respectively. Each composite was dissolved with the biomass in 100 mL of distilled water at 298 K at pH = 10, under stirring for two hours. The size of the fibers used was chosen after sieving using an electric sifter and after studying the effect of fiber grain size (40, 80, 125 and 250 μm on the biosorption efficiency). Then, the product was washed several times using distilled water to remove the excess of Zn(NO3)2·6H2O which has not been fixed onto the surface of *L.C*. Thereafter, the product obtained was transferred to a sand bath to dry it at 393 K for two hours. A similar experiment was done for the synthesis of the hybrid material under AC using two zinc electrodes (1.3 <sup>×</sup> 2.5 cm2; purity 99%) immersed into the solution while stirring. The electrical mounting comprising an AC source, a voltmeter to fix the current density at 0.5 A/m<sup>2</sup> and the voltage at 15 volts.

#### *2.3. Morphological and Crystallographic Characterizations of Hybrid Materials*

Morphological and structural characterizations were done in order to characterize the hybrid material synthesized in the presence and absence of AC.

Infrared characterizations (IR) were performed using a Perkin Elmer Spectrum using KBr pellet technique in the frequency range of 4000 to 500 cm−1. Scanning electron microscopy (SEM) was performed using a JEOL JSM 5400 scanning microscope (USA) after coating them with gold using a JEOL JFC-1199E ion sputtering device (USA). Energy dispersive spectroscopy (EDS) was planned to assess the surface elemental compositions of raw and the hybrid materials *L.C* +4% Zn2<sup>+</sup> using a JEOL JSM 5400 scanning microscope. EDS was performed after coating them with gold using a JEOL JFC-1199E ion sputtering device. X-photoelectron spectroscocopy (XPS) measurements were performed at room temperature with a SPECS PHOIBOS 150 hemispherical analyzer (SPECS GmbH, Berlin, Germany) in a base pressure of 5 <sup>×</sup> 10−<sup>10</sup> mbar using monochromatic Al Kalpha radiation (1486.74 eV) as an excitation source.

#### *2.4. Quantitative and Qualitative Characterization of the Pure L.C and the Hybrid Material L.C-Zn2*<sup>+</sup>

The physicochemical characterization of the pure *L.C* and the synthesized hybrid materials was determined by the Drift method and the Boehm titration. The determination of the zero charge point pH (pHpzc) consisted of placing 0.1 g of *L.C* in 30 mL of NaCl solution (0.01 M) at different pH ranging from 2 to 12. The initial pH is obtained by adding a certain amount of NaOH or HCl (1 M). For the first 24 h, the pH must be measured and then this operation must be repeated during the second 24 h in order to register the difference in pH. By plotting the pHf = f (pHi) curve, the pHpzc corresponds to the intersection of this curve with the straight line pHf = pHi [34].

The determination of acid-basic properties by Boehm method consisted in bringing onto contact 0.5 g of each lignocellulosic material with 25 mL of one of these bases: NaHCO3, Na2CO3 and NaOH (0.1 M) for 48 h with stirring. 10 mL of each solution was back-dosed with NaOH solution (0.1 M) after acidification with an excess of 0.1 M of hydrochloric acid [35].

#### *2.5. Biosorption of MB onto the Synthesized Hybrid Material*

The experiments used for the MB removal by biosorption have been performed in batch reactor by adding 0.1 g of adsorbent (Pure *L.C*, hybrid material elaborated by precipitation) in 100 mL of MB solution (pH = 10, Ci = 10 mg/L, J = 0.5A/m2, voltage = 15 volts, T = 298K). The electrical mounting comprising an AC source and a voltmeter. All the analyses were carried out in triplicate.

After biosorption, residual concentrations were determined by UV visible spectrophotometer biochrome Libra S.22 at λ = 663 nm based on the following equation [36]. Plastic and quartz cuvettes were used for absorbance tests.

$$\text{Dye removal} \left( \% \right) = \frac{\left( \text{C}\_{\text{i}} - \text{C}\_{\text{t}} \right)}{\text{C}\_{\text{i}}} \times 100 \tag{1}$$

where Ci is the initial dye concentration (mg/L) and Ct is the dye concentration at any time (mg/L).

#### *2.6. Kinetics Studies*

The evaluation of the MB biosorption in terms of adsorbed quantity was done following the Equation (2) [37].

$$\mathbf{Q}\_{\rm t} = \frac{\mathbf{C}\_{\rm i} - \mathbf{C}\_{\rm e}}{\mathbf{m}} \times \mathbf{V} \, (\text{mg/g}) \tag{2}$$

where Qt is the adsorbed quantity at equilibrium time, Ci is the initial MB concentration (mg/L), Ce is the residual MB concentration at any time (mg/L), V is the volume of solution (L) and m is the mass of the adsorbent (g). At equilibrium, Ci is equal to Ce and Q is equal to Qe [38].

Brouers-Sotolongo (B.S) model was used to fit the experimental data of MB biosorption. The generalized kinetic equation of the Brouers-Sotolongo model has been developed to provide a universal function for the kinetics of complex systems characterized by exponential law and/or power exponent behaviors. This kinetic model unifies and generalizes previous theoretical attempts to describe what is called "fractal kinetics". The mathematical development of this model and its application are presented in Brouers and Sotolongo [39–41].

The pseudo BSf (n, α) sorption kinetics is given by Equation (3):

$$\mathbf{Q\_{t}} = \mathbf{Q\_{e}}[1 - (1 + (\mathbf{n} - 1) \left(\frac{\mathbf{t}}{\tau\_{\mathbf{c}}}\right)^{\alpha})^{\frac{-1}{n-1}}] \tag{3}$$

where, n is order of the fractional reaction, α is "fractal time" exponent.

τc: characteristic time, *Q*e: the adsorbed quantity at saturation, *Q*t: the adsorbed quantity at any time.

#### *2.7. Evaluation of the Treated Water Qualities*

#### 2.7.1. Determination of Chemical Oxygen Demand (COD) and Total Organic Carbon (TOC)

The global mineralization was determined by measuring the chemical oxygen demand (COD) by a photometric method. The % COD removal can be estimated by (Equation (4)) [37].

$$\text{COD removal} \left( \% \right) = \frac{\text{COD}\_{\text{l}} - \text{COD}\_{\text{l}}}{\text{COD}\_{\text{l}}} \times 100 \tag{4}$$

where CODi is the initial COD (g/L O2) and CODt is the residual COD at any time (g/L O2).

The total organic carbon (TOC) in the solutions of MB and industrial wastewater was measured by a TOC analyzer (HACHIL, 550-TOC-TN model, Schwerte, Germany) and the TOC removal can be estimated as follows:

$$\text{TOC removal } (\%) = \frac{\text{TOC}\_i - \text{TOC}\_t}{\text{TOC}\_i} \times 100 \tag{5}$$

where, TOCi and TOCt are respectively the initial total organic carbon and the total organic carbon obtained after fixed time t of electrolysis treatment (ppm C.O).

#### 2.7.2. Germination Tests

The phytotoxicity test consists in the determination of the inhibitory effect of the treated water on the germination and growth potential of lettuce based on the GI. All sample experiments, including reference (pure water), were tripled. The results obtained are finally expressed according to the following relation [42]:

$$\text{GI}(\%) = \frac{\text{number of seconds produced in the sample}}{\text{number of seconds produced in the reference}} \times \frac{\text{average length of root in the sample}}{\text{average root length in the reference}} \times 100\tag{6}$$

#### **3. Results and Discussion**

#### *3.1. Surface Characterizations*

#### 3.1.1. Energy-Dispersive Spectroscopy (EDS) Analysis

The zinc deposit was analyzed with EDS (Table 1). Carbon and oxygen were the major compounds of *L.C* and hybrid materials due to their chemical composition (cellulose, hemicellulose, and lignin). In addition, the appearance of the Zn atom confirmed its fixation onto the surface of *L.C*. The AC enhanced the distribution of Zn2<sup>+</sup> into the *L.C* fibers where the rates according to the mass of Zn reached about 1.19% and 0.95% respectively for the hybrid material synthesized in presence and absence of AC.

**Table 1.** Percentages by mass of the chemical elements present on the fiber surface of the raw *Lu*ff*a cylindrica* (*L.C*) and the hybrid materials *L.C* + 4% Zn2<sup>+</sup> and *L.C* + 4% Zn2<sup>+</sup> + AC.


#### 3.1.2. Infrared Spectroscopy (IR)

Figure 1 shows that the spectrum of pure *L.C* has a vibration band at 3438 cm−<sup>1</sup> which corresponds to the O-H bond present in cellulose, hemicellulose, and lignin. In the presence of different percentages of Zn2<sup>+</sup> (1%, 2%, and 4%), obtained by precipitation (Figure 1a), the band of hydroxyl groups is observed at 3435, 3430 and 3414 cm−1. The decrease in the position of these bands can be attributed to the presence of elongation vibration of the Zn-OH bond, between Zn2<sup>+</sup> and cellulose [43]. These results indicate that the hydroxyl groups have a strong interaction with Zn2<sup>+</sup>.

**Figure 1.** Spectrum of pure *L.C* and *L.C* + (1%, 2% and 4% Zn2+) synthesized (**a**) in presence of alternating current (AC) and (**b**) in the absence of AC.

The appearance of 1383 peaks at 4% *L.C*-Zn2<sup>+</sup> is probably due to symmetric Zn-O-C vibration [44]. The bands at 1032.56, 1016.77 and 1016 cm−1, which correspond to the vibration of the C-O bond, increase in intensity, which can be attributed to the appearance of the Zn-O-C bond.

Based on Figure 1b, the rising of the percentages of Zn2<sup>+</sup> added during the synthesis of hybrid materials by precipitation under AC caused the decrease of the position of the band of hydroxyl groups. This last passed from 3438 cm−<sup>1</sup> to 3430, 3427and 3420 cm<sup>−</sup>1. This decrease in position can be attributed to the formation of Zn-OH stretching vibration [45].

Furthermore, we note the appearance of the Zn-O bond is proven by the appearance of peaks at 1383, 517 and 485 cm−<sup>1</sup> [46].

The bonds located at 1032.56, 1016.77 and 1016 cm−<sup>1</sup> correspond to the vibration of the C-O bond. An increase in the intensity of these bonds was shown when adding the Zn2<sup>+</sup>. This increase can be explained by the appearance of the Zn-O-C bond.

The band located at 1383 cm−<sup>1</sup> presents the main difference between the IR spectrum obtained the hybrid materials synthesized in the presence and absence of AC. For the hybrid material synthesized under AC, the band at 1383 cm−<sup>1</sup> appears from 1% of Zn2+. While the hybrid material synthesized by precipitation this band appears only for 4 % of Zn2<sup>+</sup>. This behavior is probably due to the effect of AC on the fixation of Zn2<sup>+</sup> ions into the *L.C* fibers by the creation of Zn-O-C bonds.

#### 3.1.3. X-ray Photoelectron Spectroscopy (XPS) Analysis

XPS measurements were performed in order to clarify the main difference between the syntheses of the hybrid material with and without AC. Based on IR characterizations; Zn is strongly bound to the lignocellulosic surface by OH hydroxyl ions which are confirmed by the presence of O-Zn-O and O-Zn bonds. Figure 2 displays the main XPS results for the hybrid material *L.C* + 4% Zn2<sup>+</sup> synthesized in the presence and absence of AC. The high-resolution XPS spectra of C(1s), O(1s), Zn(2p), Zn LMM<sup>+</sup> show noticeable differences between the material synthesis by the two methods. However, neither of them showed any impurities. The C(1s) peak, depicts a chemical shift by +/−0.5 eV in comparison with the C(1s) spectra observed for the hybrid material synthesized in absence of AC (not AC) (282.5 eV) towards a relatively high binding energy positioned at 283 eV specific to C=O and O–C–O [47]. Similar behavior was observed for the O(1s) peak which is positioned at the binding energies of 529.9 and 531.2 eV for the hybrid material *L.C* + 4% Zn synthesized in absence and presence of AC, respectively. These contributions can be assigned to O2<sup>−</sup> ions in the Zn-O bonding and to O–H groups of adsorbed water molecules [48]. The four peaks observed at the binding energies of 1017.5, 1020 and 1040 and 1044.5 eV, correspond respectively to the zinc oxide (ZnO) and the spin orbit of Zn (2p3/2) and Zn (2p1/2) for the material synthesized in absence of AC [48]. In presence of AC, two main peaks located at 1021 and 1043.5 eV are assigned to the Zn(2p3/2) and Zn(2p1/2) energy levels.

**Figure 2.** Spectra of C1S, O 1s, Zn 2p and Zn LMM of various morphologies of *L.C* + 4% Zn2<sup>+</sup> synthesized in presence of AC (-) and absence of AC (not AC -).

Both of these were symmetrical and narrow, indicating the absence of the zinc oxide. Despite this difference, these peaks of the Zn(2p) detected in both hybrid materials are related to Zn-O bonding as confirmed by previous studies [49]. Furthermore, we determined a small offset followed by a big discrepancy in terms of the intensities of peaks which were higher in case of the use of AC. These findings indicated that the obtained Zn does not occur in the same oxidation state and even if they are similar in number. It can also be attributed to the depth distribution of the atoms. This increase in intensity is explained by the diffusion of atoms towards the surface which was more pronounced compared to the hybrid material synthesized without AC. This behavior might stem from the difference in the local chemical environment. It can also be directly related to the number of atoms in the corresponding chemical state, which is confirmed by the signal Zn LMM Auger region. Another possibility is the sharing of the element with its neighbor atoms which applied a crystal field on it. This can also be interpreted in terms of the asymmetry in the bond arrangement which can create extra pressure on the element—so, the binding energy required for the ejection of the electron is slightly higher than its previous case. In the light of these results, we confirm the fixation of Zn2<sup>+</sup> onto the *L.C* by the creation of Zn-O bond as proven by IR characterizations for both methods used for the synthesis of the hybrid material. However, the environment of Zn bonding is slightly different.

#### 3.1.4. Scanning Electron Microscopy (SEM) Analysis

SEM characterizations were performed to gain further insight into the distribution of Zn2<sup>+</sup> obtained from (Zn(NO3)2·6H2O on the *L.C* surface by precipitation in presence and absence of AC. Figure 3 shows the obtained micrograph of the pure *L.C* (Figure 3a), hybrid material (*L.C* + 4% Zn2+) synthesized by precipitation (Figure 3b), and hybrid material (*L.C* + 4% Zn2<sup>+</sup>) synthesized under AC (Figure 3c). *L.C* presents a rough structure and a homogeneous appearance which is formed by bonded multicellular fibers (Figure 3a). This indicates that the existence of a large number of hydroxyl groups could provide effective interaction between *L.C* and Zn2<sup>+</sup> [26,50]. Figure 3b,c refer to the hybrid material (*L.C* + 4% Zn2<sup>+</sup>) synthesized in the absence and presence of AC. Here the Zn2<sup>+</sup> deposit is well visible and well-fixed onto the *L.C* and exhibits good dispersion without significant aggregation. The use of AC has enhanced the distribution of Zn2<sup>+</sup> into the *L.C* fibers.

(**c**)

**Figure 3.** Images of (**a**) pure *L.C*, (**b**) *L.C* + 4% Zn2<sup>+</sup> synthesized in absence of AC, (**c**) *L.C* + 4% Zn2<sup>+</sup> synthesized in presence of AC.

#### 3.1.5. Surface Chemical Analysis of Pure *L.C* and Hybrid Materials

Table 2 shows the main pHpzc values obtained for the pure *L.C* and the synthesized hybrid materials. The presence of zinc increased the pHpzc which made the fibers slightly basic. In addition, pHpzc increases with increasing the percentage of zinc added.


**Table 2.** pHpzc of hybrid material synthesized in absence and presence of AC.

The pHPZC of *L.C* is equal to 7.38. This result shows that the surface of *L.C* has a positive charge at pH values below pHPZC and should, therefore, be able to adsorb anions and a negative charge at pH values above pHPZC and should, therefore, be able to adsorb cations [51].

On the other hand, according to Table 3, for the pure *L.C*, the amount of base and the total amount of acid groups are approximately equal. These results suggest that *L.C* fibers have an amphoteric character. This amorphous character confirms the result found by the pHPZC (7.38).


**Table 3.** Different functional groups obtained on the adsorbent surfaces.

The carboxylic and phenolic functions of hybrid materials synthesized under AC (*L.C* + Zn2<sup>+</sup> + AC) are lower than those synthesized in absence of AC (*L.C* + Zn2+) (0.155 <sup>±</sup> 0.001 and 0.170 <sup>±</sup> 0.002 for the carboxylic functions, 0.400 and 0.230 for the phenolic functions, respectively). These findings confirm the increase of pHpzc obtained during the synthesis of the hybrid material in presence of AC (7.96 ± 0.005) which was slightly higher than that obtained during the synthesis in its absence (7.86 <sup>±</sup> 0.012). It is apparent that the addition of Zn2<sup>+</sup> and the use of AC slightly influenced the functional groups of hybrid materials obtained. Such behavior can be explained by the increase in the number of available sites on the surface of the materials, which enhances the uptake capacity of such materials and accelerates the biosorption kinetics.

Furthermore, the basic and lactonic functions of *L.C*-Zn2<sup>+</sup> + AC are higher than those of *L.C*-Zn2<sup>+</sup>(1.920 <sup>±</sup> 0.0001 and 1.810 <sup>±</sup> 0.0007 for basic functions and 0.511 <sup>±</sup> 0.0005 and 0.430 <sup>±</sup> 0.002 for lactonic functions, respectively). According to Mahdoudi et al. (2015), the increase in basic functions is probably due to partial or total deprotonation of the active sites of the hybrid material. Therefore, the fixation of zinc onto *L.C* is accompanied by an increase of the basic functions of its surface and subsequently an increase in the pHpzc. The dissociation of the surface oxygen groups of the acidic groups (carboxylic, lactone and phenol) confers a negative charge on the surface of Luffa cylindrica. Thus, the surface acid sites are of Brönsted type [52]. With regard to the positive charge, this may result from the existence of basic oxygen groups such as pyrones or benzopyran.

#### *3.2. Kinetics of MB Bbiosorption Assisted by AC*

The MB biosorption tests were performed at an initial dye concentration of 10 mg /L, pH = 10, a mass of the adsorbent of 0.1 g and a temperature of 298 K, V = 15 volts, J = 0.5 A/m2.

Figure 1 shows that the method of elaboration of the hybrid material has a great influence on the biosorption capacity. Indeed, the adsorbed quantity of the highest MB is obtained by the hybrid material (*L.C* + 4%Zn2<sup>+</sup> + AC). As a result, the adsorbed amount of MB increased from 3.22 mg/g for pure *L.C* to 8.81 mg/g for *L.C* + 4% Zn2<sup>+</sup> and 9.84 mg/g for *L.C* + 4% Zn2<sup>+</sup> + AC. Similarly, the MB removal rates reached about 98.4, 88.1 and 31.4% respectively for the hybrid material *L.C* + 4%Zn2<sup>+</sup> synthesized in the presence and absence of AC and the pure *L.C* (Figure 4).

**Figure 4.** Kinetics of retention of methylene blue (MB) onto different materials used *L.C*; pH = 10, Ci = 10 mg/L, m = 0.1 g, T = 298 K. (**a**) hybrid material *L.C* + 4% Zn2<sup>+</sup> synthesized in presence of AC, (**b**) hybrid material *L.C* + 4% Zn2<sup>+</sup> synthesized in absence of AC, (**c**) % color removal using different materials.

The use of AC in the synthesis of the hybrid material for different percentages of Zn2<sup>+</sup> improved the quantities adsorbed as well as the time required for the process. This behavior can be explained by the facility of the access of pollutants into the biosorbent pores and the availability of more active sites for biosorption offered by the synthesized hybrid materials.

Contrariwise, the best results were obtained for hybrid materials *L.C* + 4% Zn2<sup>+</sup> synthesized in the presence and absence of AC as the Qmax reached, respectively, 9.83 mg/g and 8.81 mg/g. Interestingly, the high amount of added Zn2<sup>+</sup> assures the increase of superficial negatively charged groups obtained by lactonic function (as confirmed by Boehm titration). Therefore, this increase of negatively charged groups increased the biosorption capacity.

When comparing biosorption when using hybrid material synthesized in the presence of AC and those synthesized in its absence as well as the pure *L.C*, definitely, the speed of biosorption and the capacity uptake were greatly enhanced. While the time required for the biosorption of MB is almost lower than that required when using the pure *L.C* and hybrid materials synthesized in the absence of AC. This is probably due to the existence of the electric field which increased the speed of movement of the molecules of MB and facilitated their access into the active sites.

#### *3.3. UV Characterizations*

To forecast the main phenomena that may occur during the MB biosorption, an UV-visible characterization was performed as shown in Figure 5. The characterization was done in order to compare the initially existing bands and the new ones that may appear during the biosorption of MB assisted by AC. The initial MB spectra display three main peaks located at 293,602 and 663 nm corresponding to the chromophore (dimethylamino group) and the aromatic rings in the MB molecule [53]. After MB biosorption, a decrease in the intensity of peaks recorded in the visible range at λmax = 663 nm relative to the naphthalene group [54] was observed for all biosorbent used. No new band appeared, thus indicating that no intermediates appeared and confirming that the color removal is probably due to a biosorption and not to an oxidation process which was probably due to the azo groups. It is interesting to note that the biosorption of MB was particularly important by using the hybrid material *L.C* + 4% Zn2<sup>+</sup> synthesized by precipitation under AC (Figure 5a) compared to the pure *L.C* (Figure 5c) and hybrid material synthesized in absence of AC (Figure 5b). In addition, the biosorption was rapidly obtained and within 120 min and a steady state concentration was achieved indicating a great ability of hybrid material for dye removal. On the contrary, the MB biosorption was tardier and the removal efficiency was lower when using the pure *L.C (*Figure 5c).

**Figure 5.** Spectra of MB biosorpion assisted by AC using (**a**) hybrid material *L.C* + 4% Zn2<sup>+</sup> synthesized in presence of AC, (**b**) hybrid material *L.C* + 4% Zn2<sup>+</sup> synthesized in absence of AC and (**c**) pure *L.C*, Ci = 10 mg/L, pH = 10, m*L.C* = 0.1g, V = 0.1 L, T = 298 K.

#### *3.4. Modeling of the Biosorption of MB Assisted by AC Using Brouers Sotolongo (B.S) Model*

The biosorption mechanism of MB onto pure *L.C* and hybrid materials elaborated by precipitation in the presence and absence of AC was studied by B.S model. The surface of the biosorbent is heterogeneous. The functions are distributed randomly hence the need to use a stochastic model (B.S). This model was chosen based on the coefficient of determination (R2), the non-linear chi-square test (χ2) and the calculated Qe (Table 4 and Table S1 in the complementary information).


**Table 4.** Comparative results for the B.S model.

Table 4 shows that B.S model BSf (1, α) presents the best fit to experimental data related to MB biosorption by *L.C* and hybrid materials. Indeed, the B.S model BSf (1, α) has the highest coefficient of determination and the lowest nonlinear chi-square test. In addition, the Qe calculated by this model gives acceptable values.

The comparison of the values of τ<sup>c</sup> calculated from BSf (1, α) shows that *L.C* + 4% Zn2<sup>+</sup> + AC has the lowest value of τ<sup>c</sup> (36.83). These results indicate that the AC synthesized hybrid materials remarkably improve both the adsorption capacity and the reaction rate of pure *L.C*.

On the other hand, the fractal constant α of *L.C* is less than one—which shows that the surface of *L.C* is heterogeneous [55]. Although the fractal constant α of the hybrid materials exceeds one, the kinetic is not clearly fractal [56].

For the hybrid materials, the fractal constant α increases with the increase of the percentage of Zn2<sup>+</sup> added. The hybrid material elaborated by AC (*L.C* + 4% Zn2+) presents a higher value of α. In fact, Selmi et al. (2018) showed that the fractal constant α increases with the number of functional surface groups [57]. Therefore, the hybrid material elaborated by AC (*L.C* + 4% Zn2+) has the highest number of functional groups.

#### *3.5. Evaluation of the Elaborated Hybrid Material for MB Biosorption: A Comparative Study*

To estimate the efficiency of the hybrid material synthesized by the precipitation method in presence and absence of AC and other low-cost material widely used for dye removal, a comparative study in terms of biosorption capacity (Qm) and efficiency was done as illustrated in Table 5. Results obtained when using the hybrid material *L.C* + (1%, 2%, and 4% Zn2+) synthesized by precipitation under AC are largely higher than those obtained when using the hybrid material synthesized in absence of AC and the other low-cost materials. The Qm shifted from 3.22 to 8.81 and 9.84 mg/g, respectively, for the hybrid material *L.C* + 4% Zn2<sup>+</sup> synthesized under AC, the hybrid material *L.C* + 4% Zn2<sup>+</sup> synthesized in absence of AC and the pure *L.C.* The Qm increased greatly when adding the 4% of Zn2<sup>+</sup> compared to the pure *L.C*. These findings can be explained by the high dispersion of active sites due to the availability of a large number of sites. The modification of pure *L.C* by different zinc rates in the presence and absence of AC increased the active sites available for MB retention. Furthermore, the rapid change in current direction has enhanced the access of pollutants into the biosorbent pores—which explained the main difference between rates achieved in both cases.


**Table 5.** Comparison of biosorption characteristics between hybrid material synthesized and other studied biosorbents used for MB biosorption.

#### *3.6. Evaluation of the Synthesized Hybrid Material on the Purification of Industrial Textile Wastewater*

Some experiments were performed in order to foresee the variation of pH, COD, turbidity, TOC and processing time during the purification of industrial wastewater (Table 6). In fact, the treated wastewater is composed of sodium carbonate, caustic soda, sodium sulfate, sea salt, wetting agent, reactive dyes, chlorine, and enzymes. *L.C* + 4% Zn2<sup>+</sup> synthesized in the presence and absence of AC and the pure *L.C* were evaluated for the purification.

**Table 6.** Comparative results obtained for the purification of industrial textile wastewater.


Results show that the hybrid material (*L.C* + 4% Zn2+) synthesized under AC is more efficient when synthesized in the absence of AC and pure *L.C* for purification of industrial textile wastewater.

The % of TOC, % of COD and the % of turbidity removals obtained for the hybrid material synthesized in presence of AC were higher than those obtained by hybrid material *L.C* + 4% Zn2<sup>+</sup> synthesized in absence of AC and pure *L.C*.

Both 87.43% of COD and 92.07% of TOC were removed from the industrial textile wastewater when using the hybrid material *L.C* + 4% Zn2<sup>+</sup> synthesized in presence of AC. 52.80% of COD and 58.82% of TOC were removed from the tested industrial wastewater when using the pure *L.C*. 76.5% of COD and 80.94% of TOC were removed when using the hybrid material *L.C* + 4% Zn2<sup>+</sup> synthesized in absence of AC, for the purification of the industrial wastewater.

The same behavior was observed for the % of turbidity removal. Values reached about 60.85 for pure *L.C*, 80.94% for the hybrid material (*L.C* + 4% Zn2<sup>+</sup>) synthesized in absence of AC and 82.99%, respectively, for hybrid material (*L.C* + 4% Zn2<sup>+</sup>) synthesized in presence of AC.

While for pH, both *L.C* and hybrid materials (*L.C* + 4% Zn2<sup>+</sup>) had a significantly poor effect. This considerable fluctuation in the evaluated parameters may be attributed to the increase of the sites available for the color retention offered by the elaborated hybrid material as indicated by the Brouers-Sotolongo modeling. This analysis revealed that almost all of the parameters satisfied the Tunisian standards TN-106-02 indicating that the treated wastewater can be discharged to any receiving water body.

#### *3.7. Phytotoxicity Test*

In order to evaluate the suitability of the raw and treated industrial textile wastewater for agricultural use, a phytotoxicity test was done. The evolution of the germination index (GI), the length of root, stem and leaf of lettuce using distilled water, raw and treated industrial textile wastewater was evaluated. The most used model for weeds is lettuce "*Lactuca Sativa*". It has been used extensively thanks to its fast germination and high sensitivity for a large variety of pollutants. Furthermore, it is often used for the examination of plants interaction in an aquatic environment [62,63].

Figure 6 (Figure S1 in the complementary information) presents the main results obtained. The germination index (GI) of the initial concentrations of the tested industrial wastewater was equal to 29.87%. After biosorption assisted by AC, the percentage of the germination index increased indicating a decrease in the toxicity of the wastewater tested. It reached respectively 72.33, 59.79, and 52.14%, respectively, when using the hybrid material *(L.C* + 4% Zn2+) synthesized in presence of AC, the hybrid material *L.C* + 4% Zn2<sup>+</sup> synthesized in presence of AC and the pure *L.C.* Results showed that all GI were higher than 50%, indicating the purity of the treated wastewater and its suitability to agricultural use.

**Figure 6.** Evolution of the % GI after coupling biosorption with AC using pure *L.C* and hybrid material *L.C* + 4% Zn2<sup>+</sup> synthesized in presence and absence of AC.

#### *3.8. Mechanism of Biosorption of MB*

An assumption of the pathway of the fixation of Zn2<sup>+</sup> ions onto the *L.C* fibers and the possible interactions between MB and the synthesized hybrid material were predicted as shown in Figures 7 and 8.

**Figure 7.** Possible pathway for the fixation of Zn2<sup>+</sup> onto the *L.C* surface.

**Figure 8.** Possible pathway for the interactions between methylene blue and the synthesized hybrid material.

Figure 7 (Figure S2 in the complementary information) presents the possible pathway of the fixation of Zn2<sup>+</sup> ions onto the *L.C* fibers in the presence and absence of AC.

For both methods used the Zn2<sup>+</sup> ions generated from the dissolution of Zn(NO3)2 in water were fixed on the surface of the *L.C* through cellulose or hemicellulose, where oxygen is found, ensuring the creation of Zn-O bond on the biomass as confirmed by IR characterizations. The use of AC has enhanced the access of the Zn2<sup>+</sup> ions onto the fibers. Therefore, favoring the participation of all Zn2<sup>+</sup> ions presents in solution in the creation of Zn-O and Zn-O-C bonds. These findings can be confirmed by IR characterizations. The band located at 1383 cm−<sup>1</sup> appears from 1% of Zn2<sup>+</sup> for the hybrid material synthesized under AC, while it appears only for 4 % of Zn2<sup>+</sup> for the hybrid materials synthesized in absence of AC ensuring an increase in the number of sites available onto the hybrid material surface.

Figure 8 (Figure S3 in the complementary information) presents the proposal of a possible pathway of the interactions between MB and the synthesized hybrid material and the effect of alternating current on the proposal process.

Water solvates MB by making hydrogen bridges hydrogen obtained from and nitrogen obtained from MB and a polar bond between oxygen and sulfur (π–π Bond).

When alternating current intervenes, the molecules of H2O enter into agitation opening the access to the free links of the nitrogen of MB which subsequently binds to the Zn-O bond created on the biomass.

#### **4. Conclusions**

We have successfully synthesized a performing hybrid material based on *Lu*ff*a cylindrica* fibers and different percentage of zinc oxide (1%, 2%, and 4%) by an easy precipitation method under AC. The fast and efficient pollutant removals are the basic benefits of this research. Results confirmed clearly the effect of AC on the modification of acidic-basic properties of the *L.C* compared to the pure one. The addition of the different percentage of Zn2<sup>+</sup> (1%, 2%, and 4%) has increased the basicity of the surface functionality by increasing the pHpzc. Boehm titration indicated that the use of AC in the preparation of the hybrid materials has decreased the carboxylic and phenolic groups and increased the lactonic group. The presence of more lactonic groups has increased the density of the negative charge on the hybrid material surface which will play a crucial effect on the positively charged dye (MB) biosorption capacity. AC has increased the fixation and the distribution of Zn2<sup>+</sup> into the *L.C* fibers by increasing the sites' number as confirmed by B.S (1, α). Physicochemical analyses revealed a prominent decrease in the COD, the TOC and the turbidity of the treated industrial textile wastewater, complied with the Tunisian Standards TN-106-02. The treated textile wastewater can be discharged to any receiving water. Furthermore, the germination indexes are higher than 50% confirmed their suitability to agricultural use.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4441/11/7/1326/s1, Figure S1: Evolution of the % GI after coupling biosorption with AC using pure *L.C* and hybrid material *L.C* + 4% Zn2<sup>+</sup> synthesized in presence and absence of AC, Figure S2: Possible pathway for the fixation of Zn2<sup>+</sup> onto the *L.C* surface, Figure S3: Possible pathway for the interactions between methylene blue and the synthesized hybrid material, Table S1: Comparative results for the B.S model.

**Author Contributions:** Conceptualization, A.O., A.K. and M.S.; Formal analysis, A.O., A.K. and M.S.; Methodology, A.O.; resources, A.O., A.K., M.S. and R.B.; Writing—Original Draft preparation, A.O.; Writing—Review and Editing, A.O., A.K., M.S., R.B. and M.V.

**Funding:** This study was supported by 5TOI\_EWAS project funded by the European Union's Horizon 2020 Research and Innovation Program under Grant Agreement No. 692523.

**Acknowledgments:** The authors express their sincere thanks to the Laboratory of Energy and Materials (High School of Sciences and technology of Hammam Sousse)for the financial support and the Unitat de QuímicaAnalítica, Departament de Química, Centre Grup de Técniques de Separacio'en Química (GTS), Universitat Autónoma de Barcelona, Bellaterra, Spain.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Filtration of Uncharged Solutes: An Assessment of Steric E**ff**ect by Transport and Adsorption Modelling**

**Simona M. Miron 1,2, Patrick Dutournié 1,2,\* and Arnaud Ponche 1,2**


Received: 3 September 2019; Accepted: 12 October 2019; Published: 19 October 2019

**Abstract:** The major aim of this work was to understand and estimate the evolution of the membrane selectivity of neutral solutes after the filtration of protein or amino acid solutions. Classical methodologies led to the estimation of the mean pore radius, different for each filtrated neutral solute. The use of pore size distribution from nitrogen adsorption/desorption experiments enabled a good description of hydraulic and selectivity performances. The modification of the membrane hydraulic properties after the successive filtration of protein solutions revealed that the decrease is quasi linear, the same for all the studied membranes and independent of prior tests. According to the experimental observations, an adsorption model was developed, considering a layer by layer adsorption in the larger pores of the membrane. The predictive obtained results are in good agreement with the experimental rejection rates, validating the assumptions.

**Keywords:** protein adsorption; neutral solute; ultrafiltration; selectivity modelling; pore size distribution

#### **1. Introduction**

Membrane-based technology offers a reliable option in a growing market to the standard separation processes (especially in terms of power consumption, addition of chemical reagents, and operating convenience) [1]. Currently, membrane separation technology is widely used to separate, concentrate, and rectify solutions for various purposes (for example, waste treatment, water desalination, purification of pharmaceuticals) in many industrial sectors (food, medicine, waste decontamination, chemical, textile industries) [2–5]. Specifically, they are used for the separation of neutral solutes from aqueous solutions (proteins for food industry, vitamins, peptides, drugs in general for the medicine industry) [6,7].

Alternatively, neutral solute filtrations are mainly used for characterizing membranes (steric effect, cut-off). For this purpose, a transport model is used. For example, Schönherr et al. [8] filtrated raffinose to estimate the mean pore radius by using the Paganelli and Solomon model [9]. Schaerp et al. [10] used the Spiegler–Kedem equation [11] derived from irreversible thermodynamics to numerically approximate the experimental rejection curves of galactose–, maltose–, and raffinose–water solutions. The reflection coefficient is expressed by the steric hindrance pore model. Ito et al. [12] filtrated five dextran samples (molecular weights ranging from 10.5 kg mol−<sup>1</sup> to 2000 kg mol−1) to investigate the modification of pore size as a function of the Ba2<sup>+</sup> concentration. Velicangil and Howell [13] estimated the steric properties of ultrafiltration membranes by using the orifice model with rejection curves of three protein solutions (papain, Bovine Serum Albumin (BSA), and ovalbumin). In recent years, several works investigated the steric effect by using the Nernst–Planck model for uncharged solutes [14]. They studied the rejection rates of vitamin B12, raffinose, sucrose, and glucose and compared the estimated mean pore radius with the results from Atomic Force Microscopy (AFM). Significant differences were observed between the estimated mean pore radius, especially regarding

pore size distribution estimated via AFM investigations. This methodology was used in many current works [15–17].

One major observation of these works is that results obtained with several molecules are different. Indeed, the filtration of different uncharged solutes leads to various estimations of mean pore radius. Another major problem, mainly related in the literature, is the adsorption of uncharged solutes in the membrane pore, and so, the partial clogging of the active layer. Proteins are indeed known to adsorb on various surfaces due to their high propensity of deformability and conformation modification, as demonstrated by Karlson et al. [18] and Norde [19]. Even if the amount of protein on hydrophobic substrates is generally high, the outer layer of membranes constituting of oxides can also absorb a non-negligible quantity of protein. Thereby, it reduces the apparent pore size and also modifies the surface properties of the active layer, as demonstrated by Huang et al. [20] and Ponche et al. [21].

For example, Robertson and Zydney [22] observed protein adsorption in ultrafiltration membrane pores, reducing the hydraulic permeability and increasing the selectivity. They estimated the decrease in pore size due to adsorption, a decrease compatible with a monolayer adsorption. A.M. Comerton et al. [23] studied the membrane adsorption of pharmaceutically active compounds. They observed that ultrafiltration (UF) membranes are more subject to adsorption than nanofiltration and reverse osmosis membranes. This phenomenon can be easily explained by the pore size of UF membranes. In most cases, the adsorption of uncharged solutes increases the membrane selectivity and decreases the hydraulic performances and, in some cases, can extend to complete pore clogging [24]. If studies referring to molecule adsorption in membrane pore are abundant [25], fewer studies deal with adsorption mechanisms in the pore, except the classical theoretical models (monolayer) [26] or for microfiltration membranes [27]. In any case, there has not been a study that dealt with the numerical estimation of solute rejection rate (predictive) by taking into account pore size distribution and pore adsorption.

In this study, the filtration of neutral solutes is performed with a ceramic ultrafiltration membrane to investigate size selectivity. To this end, the experimental pore size distribution estimated from nitrogen adsorption/desorption experiments is used as an input of the model. Adsorption phenomena are taken into account to understand the change in hydraulic performances and the membrane selectivity.

#### **2. Materials and Methods**

Experiments were performed with tubular ceramic membranes provided by TAMI Industries (Nyons, France). The active layer in TiO2 was deposited on the internal surface of an alumina macroporous tube (1 kDa, length = 25 cm, inner diameter = 8 mm). The TiO2 layer was observed by scanning electron microscopy (Philips XL30 First FEG, SEMTech Solutions, North Billerica, MA, USA) to estimate layer thickness.

Nitrogen gas adsorption–desorption isotherms were performed with a Micromeritics ASAP 2420 (Micrometrics, Ottawa, Canada, apparatus at T = 77 K. Prior to experiments, each sample (broken pieces of TiO2 membrane and alumina tube) was out gassed to a residual pressure lower than 0.8 Pa at 350 ◦C for 15 h. The micropore volume was calculated using the *t*-plot method. The cumulative volume and the pore size diameter (distribution) were calculated using the density functional theory method (DFT).

Filtration tests were performed in a laboratory pilot plant (stainless steel), previously described in literature [28]. The studied solution was stored in a 5 L tank and a volumetric pump provided solution circulation for tangential flow filtration (retentate). The flow was controlled by an analog flow rate sensor. The experiments were performed at 700 L/h, a value corresponding to a mean fluid velocity higher than 5 m/s to avoid concentration polarization at the surface of the active layer. After filtration, the permeate flow was sampled for analysis. Both retentate and permeate returned into the solution tank. The applied pressure was adjusted by a manual valve (4 to 12 bar) and was measured by two sensors upstream and downstream the membrane carter. A cooling unit maintained

the fluid temperature at 25 ◦C. Between each filtration test, the experimental set-up was rinsed with demineralized water (conductivity < 0.1 μS/cm).

Before performing filtration tests, a conditioning step of the membrane was required. This step consisted of the filtration of pure water until hydraulic performances were in steady state.

Filtration tests of uncharged solute-water solution ware performed with vitamin B12 (Alfa Aesar, purity 98%), *L*-phenylalanine (Fluka, purity 99%), *L*-tyrosine (Fluka, purity 99%), and lysozyme (Sigma-Aldrich, from chicken egg white, activity > 70,000 U/mg). The concentrations of retentate (*Cr*) and permeate (*Cp*) samples were investigated by absorbance measurements with a UV-visible spectrophotometer (Lambda 35, Perkin Elmer Instrument). Information about the studied molecules are given in Table 1.

**Table 1.** Information on the molecules used for experimental tests. **Molar Mass**


The observed rejection rate *R* was calculated using Equation (1):

$$R = \frac{\mathbf{C}\_r - \mathbf{C}\_p}{\mathbf{C}\_r}.\tag{1}$$

Several tests of analytical measurement uncertainties were performed (repeated eight times). The maximal error was 1.9% for retentate and 2.9% for permeate samples. The rejection rate uncertainty was Δ*R* = (1 − *R*) 1 *Cp* <sup>Δ</sup>*Cp* <sup>+</sup> <sup>1</sup> *Cr* Δ*Cr* and, in all cases, was inferior to 4.8% (result for a rejection rate close to 0).

During the experimental tests, the hydraulic properties were monitored. To do this, between each experiment test, a pure water filtration test was performed. The permeate flux (*Jv*) was measured and plotted for different applied pressures (Δ*P*). The hydraulic permeability (*Lp*) was obtained from the slope of the linear curve via Equation (2):

$$Jv = \frac{L\_p}{\mu} \Delta P.\tag{2}$$

Three series of tests were performed with the same membrane. Between each series the membrane was regenerated. The regeneration consisted of a hydrothermal treatment (five days in water at 105 ◦C) to recover its original hydraulic properties.

Filtration tests of neutral solutes were used to investigate size selectivity. Two models were studied: model A assumed a uniform one size distribution (average pore radius) and the second (model B) used an experimental size distribution determined from nitrogen adsorption/desorption experiment as an input.

In model A, the mean pore radius was estimated by numerically approximating rejection rates of the studied solutes. The equation used to approximate the experimental results was the solution of the mass balance (differential equation) in the membrane pore. The differential equation expressing solute mass balance is described by the Nernst–Planck approach for uncharged solutes [29,30], assuming uniform dispersed, one size cylindrical pores. This differential equation was solved with the equality of chemical potentials on both sides of the active layer [31] at the pore/solution interface. Starting from these assumptions, the rejection rate of an uncharged solute [32] can be calculated with Equation (3):

$$R = 1 - \frac{qK\_c}{1 - (1 - qK\_c) \exp\left(-\frac{K\_c r\_p^{-2} \Lambda P}{8\mu K\_d D\_{\rm ov}}\right)}\tag{3}$$

With ϕ = <sup>1</sup> <sup>−</sup> *<sup>r</sup> rp* 2 the steric partitioning coefficient.

The second model (model B) implied that we had an experimental distribution of pore size. In this case, the permeation flow rate was assumed to be the sum of the flow rate of each pore (Equation (4)). The flow rate in the pore was calculated using Hagen Poiseuille's law:

$$Q = J\_{\upsilon} S\_{\mathfrak{m}} = \frac{Lp}{\mu} \Delta P \times S\_{\mathfrak{m}} = \sum\_{i} np(i) \times q(i) = \sum\_{i} np(i) \times \frac{\pi r\_{p}(i)^{4}}{8\mu \Delta \mathbf{x}} \Delta P. \tag{4}$$

The rejection rate *R* was calculated using the Nernst–Planck approach for uncharged solutes (Equation (3)). The calculation was carried out for each pore size, summing all the contribution weighting by the pore number and the relative flow rate (Equation (5)) as follows:

$$R = \frac{1}{Q} \sum\_{i} np(i).q(i).\mathcal{R}(i). \tag{5}$$

#### **3. Results**

#### *3.1. Experimental Results and Estimation of Steric E*ff*ect*

Experimental tests were performed to understand the physical phenomena that act on the mass transfer mechanisms in a porous media. For this purpose, the monitoring of hydraulic and selectivity properties was required. Three series of tests were performed with the same membrane. Between each series of tests, the membrane was regenerated according to the protocol previously described in the Materials and Methods section. This regeneration aimed to recover the initial membrane properties. Table 2 provides experimental results obtained for three series in terms of maximal rejection rate (selectivity property) and hydraulic permeability (hydraulic performance). For each series, the results are in chronological order. These series were chosen as the first molecule filtrated was different (*L*-phenylalanine for series 1, vitamin B12 and Lysozyme for series 2, and *L*-Tyrosine for the last).

**Table 2.** Filtration experiments (chronologic order) for the three series of tests.


The first experiments (left-hand column) were performed with a new membrane. After conditioning, the hydraulic permeability was 6.7 <sup>×</sup> <sup>10</sup>−<sup>14</sup> m3 <sup>m</sup><sup>−</sup>2. The four studied molecules were filtrated one after another, from the smallest to the largest. The rejection rate increased as filtration tests were performed and, at the same time, the hydraulic permeability significantly decreased, suggesting that a part of the protein was adsorbed at the membrane surface (or in the pore), reducing mass transfer. Nevertheless, filtration of VB12 does not seem to modify the membrane hydraulic performances, and, thus, was chosen as the model solute to follow the membrane selectivity properties. The solute rejection rates obtained with the membrane still having its initial properties (i.e., membrane that had only filtrated water and vitamin B12 after regeneration) were 5% for *L*-phenylalanine (test 1 series 1), 5% for *L*-Tyrosine (test 1—series 3), 20% for Vitamin B12 (test 1 series 2), and 65% for Lysozyme (test 2—series 2). These rejection rates are in good agreement with the size of the studied molecules, but much lower than expected for a membrane with a cut-off of 1 kDa.

After these first experimental tests, the membrane permeability and the solute transmission decreased. This behavior was illustrated by the rejection rate of vitamin B12, which increased as protein or amino acid solutions were filtrated. For example, it increased from 20% to 90% after several filtration tests of lysozyme solutions (series 2), indicating that steric effect increased significantly. This can be explained by protein or amino acid adsorption at the membrane surface and or in the pore, thereby restricting the transfer of solute through the porous medium. These adsorption phenomena are not completely irreversible, since after the regeneration treatment the membrane recovered its initial properties.

The estimated mean pore radii are given in Table 2 for series 1 (from Equation (3)). The results (calculated average pore radii) corroborate previous observations. Indeed, the average pore radius decreased as protein filtration was performed. The results obtained with the different solutes show that this current model (model A) remains unsatisfactory for estimating steric effect because the results are dependent on the studied molecule. The average pore radius estimated with vitamin B12 was systematically lower than for lysozyme and *L*-phenylalanine. Bowen et al. [14] obtained the same results for the filtration of four uncharged solutes (vitamin B12, raffinose, sucrose, and glucose). Indeed, the estimated mean pore radius can vary by a ratio of 2:1. They also observed a modification of water permeability after filtration of each solute.

Additional tests were performed with mixtures of uncharged solutes (vitamin B12 and lysozyme) in water or salted water. These tests were performed following the tests of series 1.

The results (Table 3) show that the rejection rate of the solute was not modified by any another solute in the solution or by salt (NaCl—5 mM) in dilute solution. In the present case, it seems that the solutes in solution did not interact between them. However, in the literature, several studies [33–35] showed an increase in neutral solute transmission in saline solutions. This phenomenon is explained by a partial dehydratation of the molecule in the membrane pore, facilitating the transport through the membrane active layer. This behavior is specifically observed in nanofiltration. In the present study, the filtrated solutes, the porous material (membrane active layer), and the pore size were different, which could explain the observed differences.


**Table 3.** Filtration tests of solute mixture.

Moreover, the difference in estimated mean pore radius according to the studied molecule was confirmed. There are two possible explanations. First, the one size cylindrical pore hypothesis is a poor and inadequate representation of the porous medium. Second, the solute interacts with the membrane surface according to the chemical groups of both the solute and the active layer (van der Waals forces or acid-base interactions). To investigate the first possibility, a pore size distribution was studied. To this end, nitrogen adsorption/desorption experiments were used to measure the porous volume according to the pore size.

Figure 1 shows the cumulative porous volume obtained during the nitrogen adsorption step according to pore size. This curve can be suitably approximated by a log-normal distribution. These results were substantiated by Scanning Electron Microscopy (SEM) investigations. Indeed, the SEM images (Figure 2c) show a top view of the active layer. The surface constituted sintered titania aggregates separated by nanometric cracks or channels. The larger ones, measuring about 4–10 nm, were of a lower amount than the smaller ones (less than 1 nm wide).

From these experimental results, we assumed that the porous medium was constituted of cylindrical and unidirectional pores according to the previous size distribution (Figure 1) and calculated the pore number as a function of the pore size.

Assuming that the flow rate in each pore can be described by the Hagen–Poiseuille law, the permeation flux can be estimated with Equation (4). From these results, it is possible to estimate the active layer thickness by equaling the observed permeation flux with the flow rate calculated by Equation (4). For this purpose, the studied range of pore diameter (0–14 nm) was divided into sub-ranges of pore size from *i* = 0 to 1400. The active layer thickness was around 2.9 μm. Previous observations [28] performed by microscopy (SEM) showed that the active layer was about 1–2 μm thick. Figure 2a shows a cross section of the membrane obtained after breaking it. The different alumina layers (different porosities) are visible below the titania active layer (full grey layer). The apparent thickness (Figure 2c) of this layer is in agreement with previous investigation.

Calculations were performed for modelling selectivity performances of the new or regenerated membranes for the four studied molecules (i.e., for membranes, which have only been in contact with water). These calculated results were compared with the experimental observed rejection rates of the four studied solutes (Table 4). The results are very close to the experimental ones for the four studied molecules. This comparison indicates that taking into account pore size distribution provides a good way to estimate rejection rates of neutral solutes.

**Figure 1.** Cumulative porous volume (red curve) versus pore diameter and pore size distribution (blue curve).

**Figure 2.** SEM photographs of (**a**,**b**) cross section at different enlargements and (**c**) top view of the active layer.

**Table 4.** Experimental and calculated (pore size distribution) rejection rate of solutes (membrane with initial properties).


#### *3.2. Hydraulic Performance Loss and Adsorption in the Pore*

The hydraulic performances of the membranes decreased as filtration tests of neutral solutes were performed (except for Vitamin B12). At the same time, selectivity performances increased, indicating an increase in the steric effect limiting the mass transfer. As shown in Table 2 for the first series, the average pore radius (calculation relative to each filtrated molecule) decreased, step by step, until a minimum value. Four clogging/blocking mechanisms are classically reported in the literature: complete blocking (the filtrated molecule blocks the pore inlet), standard blocking (adsorption of the molecules in the pore), intermediate blocking (formation of a non-continuous layer, blocking partially the flow inlet), and a cake filtration. This last one often occurs in dead-end filtration. In our case, it can be ruled out owing to the tangential flow and associated shear stress. The intermediate blocking cannot explain both permeate flux decrease and selectivity increase. Indeed, to that end, non-continuous layers should preferentially grow, clogging the larger pore inlet only. The first mechanism, i.e., pore blocking, cannot be a plausible hypothesis, because during filtration of amino acids, the pores that clog should be the smallest, which would result in a decrease in rejection rate. So, the only plausible explanation is protein or amino acid adsorption in the larger pores, reducing the flow rate and increasing the

selectivity. F. Wand et al. [36] observed that standard blocking occurred first during ultrafiltration of colloid–water solutions (dead-end filtration experiments). K. Katsoufidou et al. [37] observed a rapid irreversible membrane fouling during ultrafiltration of humic acid due to internal pore adsorption.

Figure 3 shows the hydraulic permeability of the membrane divided by the one obtained just before the first filtration test of lysozyme for series 1 and 3 (presented in Table 2). These results were compared with two test series performed in the same operating conditions with another membrane (series 4 and 5). The hydraulic permeability decreased as filtration tests were carried out (quasi linear behavior), indicating the same adsorption kinetics, regardless of the studied membrane and its past experiments. Assuming one size pore distribution, the hydraulic permeability was proportional to the fourth power of the pore radius (Poiseuille flow). In these conditions, the pore radius estimated by the filtration tests of a neutral solute should vary according to the fourth-root dependence of the number of tests.

**Figure 3.** Relative hydraulic permeability versus the number of filtration tests of lysozyme performed.

Figure 4 compares the estimated mean pore radius for vitamin B12 and Lysozyme filtration tests (series 1) with the mean pore radius calculated using a linear function of the fourth-root of the cumulative number of lysozyme test. For the two cases, the estimated pore radius can be approximated by *rp*(*i*) = *rp*(0) × (1 − 0.165*i*) 0.25.

The same investigation with pore size distribution was not possible owing to the number of adjustable parameters (i.e., *Np*(*i*) and *rp*(*i*)).

Steric hindrance after protein adsorption is required in order to predict the selectivity performances of the membrane. Classical models of adsorption or adsorption kinetics do not provide information about flow restriction in pores.

**Figure 4.** Mean pore radius estimated from filtration experiments and calculation of neutral solute (series 1: Vitamin B12 and Lysozyme).

In these conditions, to model these adsorption phenomena, we considered the adsorption of spherical molecules to be by uniform layers in the pore, reducing its radius of two stokes radius of the adsorbed molecule. This assumption is an arbitrary but required hypothesis for calculating the pore size distribution after molecule adsorption. The active layer thickness and the total number of pores were assumed to be unchanged (only their sizes can be modified). The calculation was carried out in all the pores with diameters twice larger than the molecule size. This procedure was numerically repeated until the calculated mass flow rate equaled the experimental permeation flux. When the equality of experimental and numerical hydraulic performances was reached, the model became predictive and capable of calculating the rejection rates of other filtrated molecules (i.e., Vitamin B12, *L*-Tyrosine, *L*-Phenylalanine, and Lysozyme).

Figure 5 shows the pore size distribution of the membrane after several filtration tests of lysozyme solution (series 1) compared to the initial one. This pore size distribution was obtained by equaling the experimental and calculated permeation fluxes. The results show that adsorption phenomena took place in the larger pores, partially clogging them and reducing their apparent diameters. Consequently, the pores with an initial size higher than 3.9 nm (corresponding to pores larger than adsorbed solute) completely disappeared and the pore quantity with a diameter in between the 1.8–2.8 nm range significantly increased.

To validate this modeling, rejection rates of the four filtrated molecules were calculated before and after lysozyme adsorption.

Experimental and calculated rejection rates of the studied molecules are given in Table 5 for the membrane before (initial properties) and after several filtration tests of lysozyme solutions. Taking into account that a pore size distribution provides a good description of membrane selectivity performances and its modification over time, from this new size distribution, the rejection rates of *L*-phenylalanine, *L*-Tyrosine, and Lysozyme increased and are in good agreement with the experimental results. Nevertheless, the rejection rate of vitamin B12 increased up to 60%, which is rather different

regarding the experimental results (between 85% and 90%). These results are basically prior image and need to be further investigated and confirmed by other membranes and molecules.

**Figure 5.** Initial and calculated pore size distribution after lysozyme filtration tests (series 1).


#### **4. Conclusions**

In this work, filtration tests of four uncharged solutes were performed in order to investigate the steric exclusion effects. These tests were performed with new or regenerated membrane to investigate the pore size distribution and selectivity of the membrane regarding each solute. Taking into account a pore size distribution experimentally determined with nitrogen adsorption/desorption techniques made it possible to accurately describe the membrane selectivity performances. The simulated and experimental rejection rates of the four studied molecules are in good agreement for new membranes. Successive filtration tests showed that proteins and amino acids interact with the membrane (adsorption in the larger pore), reducing the hydraulic performances and increasing the selectivity. To describe this phenomenon, the pore size distribution was recalculated by considering molecule adsorption in the larger pore. In this way, the calculation was performed by equalizing experimental and numerical permeation flux. The simulated results provide a good perspective of trends in membrane selectivity for neutral solute filtration.

**Author Contributions:** Conceptualization, P.D. and A.P.; Methodology, S.M.M., P.D. and A.P.; Validation, S.M.M., P.D. and A.P.; Modelling P.D.; Investigation, S.M.M., P.D. and A.P.; Writing—Original Draft Preparation, S.M.M., P.D. and A.P.; Supervision, P.D. and A.P.

**Funding:** This research received no external funding

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Glossary**


#### **Greek Letters**


#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Use of Lignite as a Low-Cost Material for Cadmium and Copper Removal from Aqueous Solutions: Assessment of Adsorption Characteristics and Exploration of Involved Mechanisms**

**Salah Jellali 1,\*, Ahmed Amine Azzaz 2, Mejdi Jeguirim 2, Helmi Hamdi <sup>3</sup> and Ammar Mlayah <sup>4</sup>**


**Abstract:** Lignite, as an available and low-cost material, was tested for cadmium (Cd) and copper (Cu) removal from aqueous solutions under various static experimental conditions. Experimental results showed that the removal efficiency of both metals was improved by increasing their initial concentrations, adsorbent dosage and aqueous pH values. The adsorption kinetic was very rapid for Cd since about 78% of the totally adsorbed amounts were removed after a contact time of only 1 min. For Cd and Cu, the kinetic and isothermal data were well fitted with pseudo-second order and Freundlich models, respectively, which suggests that Cd/Cu removal by lignite occurs heterogeneously on multilayers surfaces. The maximum Langmuir's adsorption capacities of Cd and Cu were assessed to 38.0 and 21.4 mg g−<sup>1</sup> and are relatively important compared to some other lignites and raw natural materials. Results of proximate, scanning electron microscopy/energy dispersive X-ray spectroscopy (SEM/EDS), Fourier transform infrared spectroscopy (FTIR) and X-Ray diffraction (XRD) showed that the removal of these metals occurs most likely through a combination of cation exchange and complexation with specific functional groups. The relatively high adsorption capacity of the used lignite promotes its future use as a low cost material for Cd and Cu removal from effluents, and possibly for other heavy metals or groups of pollutants.

**Keywords:** lignite; heavy metals; adsorption; batch; isotherm; mechanism

#### **1. Introduction**

The contamination of water resources by heavy metals contained in industrial effluents is an important worldwide concern due to their toxicity, low biodegradability and high accumulation capacity in water-living organisms [1]. When transferred to humans through the food chain, these metals could be accumulated in various body organs and tissues causing life-threating illness and possible damages to vital systems [1]. Cadmium and copper are among the most toxic heavy metals. Indeed, exposure to cadmium, mainly used in batteries and the coating and plating industry, could result in stomach pains, bone fracture, possible infertility as well as serious damages to the central nervous and immune systems [2]. High ingested amounts of copper could increase infection frequencies, cardiovascular risks, and alterations in cholesterol metabolism [3]. The removal of these pollutants from industrial wastewater before discharge into the environment is, therefore, an urgent task to be appropriately achieved.

Many technologies have been developed for heavy metal removal from industrial effluents. They include chemical precipitation, reduction, electrodialysis, and membrane

**Citation:** Jellali, S.; Azzaz, A.A.; Jeguirim, M.; Hamdi, H.; Mlayah, A. Use of Lignite as a Low-Cost Material for Cadmium and Copper Removal from Aqueous Solutions: Assessment of Adsorption Characteristics and Exploration of Involved Mechanisms. *Water* **2021**, *13*, 164. https://doi.org/ 10.3390/w13020164

Received: 13 December 2020 Accepted: 6 January 2021 Published: 12 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

separation [4–7]. These technologies have shown interesting removal efficiencies at laboratory scale. However, their upscaling to real cases has been hindered by several limitations such as high capital and exploitation costs, very sensitive operational conditions, the use of large amounts of chemicals, and the production of secondary sludge that has to be sustainably handled [7].

The adsorption technique has been pointed out as a promising, eco-friendly and low-cost method for heavy metal removal from aqueous solutions [8–12]. In general, adsorbents with high pH, important specific surface area, well developed microporosity and that are rich in specific functional groups are used for the efficient removal of heavy metals form effluents. Once the adsorption process is completed, the loaded-adsorbents could be regenerated and heavy metals could be recovered for a further subsequent use with respect to the circular economy and sustainable development concepts [13,14]. Various mineral and organic materials have been tested for Cd(II) and Cu(II) removal from synthetic/real wastewaters. They include powdered marble wastes [15], clays [16], lignocellulosic biomasses [17], biochars [18,19], activated carbons [20,21] etc.

Lignite, often referred as brown coal, is a sedimentary rock that is naturally formed from naturally compressed peat. It is a low cost material and available in various countries at high amounts [22]. Besides its classical use for energy generation, raw lignite has been tested for the removal of various pollutants from aqueous solutions such as phenol and chlorophenol [23], trichloroethylene [24], phosphorus [25] and ammonium [26]. A special focus has been dedicated to the use of these raw materials as efficient adsorbents for heavy metals form aqueous solutions [27–30]. As such, these materials have relatively high contents of functional groups such as carboxyl, alcoholic and carbonyl groups that could complex with metal ions [27,31]. Furthermore, they have a large cation exchange capacity allowing cationic pollutants to be adsorbed [32].

Laboratory investigations regarding heavy metal removal by raw lignite have demonstrated that it could be considered as an interesting candidate for single metal removal form aqueous solutions [27,30,32]. However, when in multicomponent systems, the competition phenomenon decreases their maximal adsorption capacities. In both systems, results are sometimes contradictory and highlight that the ability of heavy metal retention depends on their physico-chemical characteristics as well as the lignite properties [33,34]. For instance, Pehlivan et al. [34] studied the adsorption of Pb, Cd, Cu, Ni, and Zn on several Turkish lignite materials. They showed that the order of metal adsorption ability depends on the lignite type. However, Pb and Cu were the most adsorbed components in contrast to Ni and Zn. Furthermore, Pentari et al. [30] studied the removal of Pb, Cd, Zn, and Cu from aqueous solutions by a lignite from Greece. They found that in single mode, the ability adsorption order was as follows: Pb > Cd > Cu > Zn. In a multicomponent system, they showed that Cu sorption was the most affected, while Zn adsorption capacity remained quasi-constant. For the same metals in a multicomponent system, Doskoˇcil and Pekaˇr [33] found a different order: Pb > Cu > Zn > Cd. On the other hand, the involved mechanisms during heavy metal adsorption onto raw lignite are still not well identified. An in-depth kinetic and isothermal modeling study combined with lignite analysis before and after metal adsorption by using advanced techniques could reduce this gap.

As a consequence, the principal objectives of this work were: (i) to assess the adsorption characteristics of Cd and Cu onto a Tunisian lignite under various experimental conditions including contact time, initial concentration, pH, adsorbent dosage, and competition with other metals, (ii) to compare the efficiency of this lignite with raw and modified lignite available in the literature, (iii) to better understand the involved adsorption mechanisms of Cd and Cu through the combination of an in-depth modeling study and various physico-chemical analyses.

#### **2. Materials and Methods**

#### *2.1. Adsorbent Preparation and Characterization*

Raw lignite was collected from the Cap Bon region (northeastern part of Tunisia). This lignite was used in its natural state after a drying step at 60 ◦C for 24 h followed by manual grinding in ceramic grinder. The fraction with diameter size lower than 63 μm was selected for the adsorption tests. The preliminary characterization of the used lignite included the determination of its: (i) mineral composition by X-ray fluorescence spectrophotometer (Philips, Eindhoven, The Netherlands), (ii) Brunauer–Emmett–Teller BET specific surface area through N2 gas adsorption method using a gas adsorption analyzer (Quantachrom Autosorb 1 sorptiometr), and (iii) pH of zero-point-charge value (pHZPC) [35].

Furthermore, advanced analyses of the lignite before and after metals adsorption were performed for a better identification of the involved mechanisms. They included the assessment of: (i) the morphology and qualitative composition through scanning electron microscopy (SEM) coupled with energy dispersive X-ray (EDX) (Philips model FEI model Quanta 400 apparatus, Amsterdam, The Netherlands), (ii) proximate analysis using a TGA/DSC3+ device (Mettler-Toledo, Greifense, Switzerland), and (iii) the existing functional groups through a Fourier transform infrared (FTIR) analysis using an Equinox 55 spectrometer (Brucker, Billerica, MA, USA). The FTIR spectra were assessed between 4000 and 400 cm−<sup>1</sup> with a resolution of 1 cm<sup>−</sup>1. Experimental protocols for the above-cited analyses have already been detailed in a previous paper [36].

#### *2.2. Synthetic Heavy Metals Solutions Preparation and Analysis*

Cadmium nitrate (Cd(NO3)2), copper nitrate (Cu(NO3)2), lead nitrate (Pb(NO3)2), and zinc nitrate (Zn(NO3)2) were used for the preparation of four stock solutions at a concentration of1gL−<sup>1</sup> each (Fisher Scientific, Waltham, MA, USA). These solutions were used throughout this study for the preparation of adsorption solutions at precise concentrations. Metal concentrations were measured thorough an atomic absorption spectrometer (AAS) with an air-acetylene flame (Perkin Elmer Analyst 200, Waltham, MA, USA). The wavelengths used for the analysis of the Cd, Cu, Pb and Zn were 228.8, 324.8, 283.3, and 213.9 nm, respectively. The initial pH values of the solutions were adjusted by using dilute sodium hydroxide or nitric acid.

#### *2.3. Batch Adsorption Investigation*

Cd and Cu removal efficiency from aqueous solutions by raw lignite was carried under static conditions (batch mode). It consists in shaking, at room temperature (20 ± 2 ◦C), a given mass of lignite in 50 mL of aqueous solution containing the metal at a fixed concentration for a desired contact time at 400 rpm by using a Variomag-poly15 magnetic stirrer. Then, the suspension was filtrated through 0.45 μm cellulose acetate filter before analysis with AAS. During this study, the effect of the following experimental conditions on Cd and Cu removal efficiency were assessed: (i) particle size distribution for four lignite granulometries: <63 μm; between 63 and 500 μm; 500–1000 μm, and 1000–2000 μm; (ii) the contact time of 1; 5; 10; 20; 40; 60, and 90 min; (iii) the initial aqueous pH for 2.0; 3.0; 4.0 and 5.0, and (iv) lignite dosages for 0.4; 1.0; 1.6; 2.0; 2.4; 3; 3.5 and 4.0 g L−1. During these assays, the default following parameters were used: a lignite size fraction lower than 63 μm, a contact time of 90 min, an initial Cd or Cu concentration of 100 mg L<sup>−</sup>1, an initial pH of 5 and a lignite dosage equal to 2 g L<sup>−</sup>1. Finally, the competing effect was determined for a multicomponent solution containing Cd, Cu, Pb and Zn at constant concentrations of 30 mg L−1. All these experiments were carried out in triplicate and mean values are reported in this work. The standard deviation for all assays was lower than 5%.

The adsorbed metal amount at a given moment '*t*', (*qt*) and the related removal yield (*Yt*) were determined as follows [37]:

$$\eta\_t = \frac{\left(\mathbf{C}\_0 - \mathbf{C}\_t\right)}{D} \tag{1}$$

$$\mathcal{Y}\_{\mathfrak{t}}(\%) = \frac{(\mathbb{C}0 - \mathbb{C}t)}{\mathbb{C}\_0} \times 100 \tag{2}$$

where *C*<sup>0</sup> and *Ct* (mg L−1) are initial metal concentrations and at a given time '*t*', respectively, and *D* is the used adsorbent dose (g L<sup>−</sup>1).

Cadmium and copper adsorption kinetics were fitted to three standard models namely, pseudo-first order (PFO), pseudo-second order (PSO), and intraparticle and film diffusion. Moreover, the experimental isothermal data were fitted to Langmuir, Freundlich and Dubinin–Radushkevich (D-R) isotherm models. The agreement between the experimental and theoretical adsorbed amounts was assessed through the determination of the average percentage errors (*APE kinetic* and *APE isotherm*) as follows:

$$APE\_{kinetic} \left(\%\right) = \frac{\sum \left| \left( q\_{t,exp} - q\_{t,theo} \right) / \left. q\_{t,exp} \right|}{N} \times 100\right. \tag{3}$$

$$APE\_{isotherm}(\%) = \frac{\sum \left| \left( q\_{\varepsilon, exp} - q\_{\varepsilon, \text{theo}} \right) / q\_{\varepsilon, exp} \right|}{N} \times 100\tag{4}$$

where *qt,exp* and *qt,theo* (mg g<sup>−</sup>1) are the experimental and the theoretical adsorbed amounts at a given time '*t*'; *qe,exp* and *qe,theo* (mg g−1) are the experimental and the theoretical adsorbed masses at equilibrium.

#### **3. Results**

#### *3.1. Lignite Characterization*

The XRF analysis of the lignite showed that it contains various minerals. Silica, Sulfur, iron, and aluminum exist at relatively high contents of 6.2%, 5.1%, 2.6%, and 2.5% (dry mass), respectively. Lower contents of 0.06%, 0.46%, 1.21%, and 0.21% were observed for Na, K, Ca, and Mg, respectively. Some of these elements could be exchanged with Pb(II) during its adsorption by lignite. On the other hand, the used material has higher concentrations of Si, Al, Fe and Na and lower Ca and Mg content than a commercial and natural lignite from Czech Republic [38,39]. The BET surface area of the lignite was assessed to 11.2 m<sup>2</sup> g−1, which is in the range of BET values reported for Greek lignites (between 3.6 and 23.8 m2 g−1) [30]. For instance, it is about 12-fold higher than a lignite from Poland (0.91 m2 g−1) [23], and about two-fold higher than a Hungarian lignite (5.3 m2 g−1) [25]. This relatively higher surface area suggests that the used lignite could exhibit more metal adsorption capacities compared to other lignites reported in the literature. On the other hand, its pHZPC was estimated to 3.6. It is lower than that of a Czech lignite (5.0) [38], and a Polish lignite (6.2) [23]. In contrast, it is higher than pHZPC of a Hungarian lignite (2.6) [25]. Since the net charge surface of the adsorbent becomes negative for aqueous pH values higher than pHZPC, the removal of positively charged metals by the current lignite through electrostatic attraction will be favored for a wide pH range.

#### *3.2. Batch Adsorption Results*

#### 3.2.1. Effect of Lignite Granulometry

The impact of lignite granulometry on Cd and Cu removal from the synthetic solution was carried out for initial metal concentrations of 100 mg L−<sup>1</sup> each, initial pH of 5 and a dosage of 2 g L−1. The experimental results (Figure 1) showed that coarser is the lignite fraction, lower is Cd or Cu removal efficiency. Accordingly, the highest adsorbed amounts of Cd (26.1 mg g−1) and Cu (18.6 mg g−1) were observed for the finest granulometry (<63 μm). These adsorbed Cd and Cu amounts decreased by about 25% and more than 65%, respectively when lignite particles of 1–2 mm size were used. This outcome could be attributed to the lower microporosity and surface area generally observed for coarser media [40]. Similar behavior was observed by PN and CP [41] when investigating an oily effluent treatment by hard wood based adsorbents. They reported a decrease of the removal efficiency by about 30% when the average size particles were increased from 0.8 to 3.5 mm.

**Figure 1.** Impact of lignite particle size on Cd and Cu removal efficiency (C0 = 100 mg L<sup>−</sup>1,D=2gL−1; t = 60 min; pH = 5; T = 20 ± 2 ◦C).

3.2.2. Effect of Contact Time—Kinetic Study

The removal of Cd and Cu by the used lignite is clearly a time-dependent process as shown in Figure 2. Indeed, their removal was very rapid at the beginning of the adsorption (especially for Cd) since about 78% and 44% of the totally removed amounts were adsorbed after only 1 min for Cd and Cu, respectively. This finding suggests that Cd removal occurs mainly through surface reactions [11]. The high reactivity of the used lignite could result in an important energy saving when such process is scaled up for field investigations. After this contact period, the adsorbed amounts continue to increase but at much slower rate. This behavior is linked to an intraparticle diffusion inside the pores of the lignite and adsorption by functional groups through complexation process [42]. The equilibrium state which corresponds to quasi-constant adsorbed amounts was reached after approximately 60 min for both metals. This time is 6-fold lower than the one reported by Havelcova et al. [38] when investigating Cd, Cu and Zn removal by a local lignite from south Moravian coal field (Czech republic) and 12-fold lower than the duration observed by Pentari et al. [43] for Cd removal by a raw and iron doped Greek lignite. However, relatively similar equilibrium contact times were observed for Cd and Cu removal by various Greek lignites [30]; Cu, Pb and Ni removal by two Turkish lignites [32]; Zn removal by a commercial lignite from South Korea [44] and Cr(VI) adsorption onto a raw Iranian lignite [28]. For economic reasons, low contact times, ensuring percentages removal of more than 80% of the totally adsorbed Cd or Cu amounts could be used in real case application. These durations correspond to only 5 and 20 min for Cd and Cu, respectively.

At equilibrium, the lignite sample used in this study removed Cd better than Cu. Indeed, the adsorbed Cd amount by was assessed to 26.1 mg g−<sup>1</sup> which is about 41% higher than Cu (Figure 2). A Similar trend was observed by various studies dealing with heavy metal adsorption onto lignite [30,38]. This behavior is mainly imputed to their physico-chemical properties, especially electronegativity, ionic potential and ionic radius [45].

**Figure 2.** Kinetics of Cd and Cu removal by lignite and its fitting with PFO and PSO models (C0 = 100 mg L<sup>−</sup>1,D=2gL−1; t = 60 min; pH = 5; T = 20 ± 2 ◦C; Exp.: Experimental value).

The parameters of the three theoretical models: PFO, PSO and diffusion models are given in Table 1. Based on these calculated parameters, it can be clearly deduced that the PFO model does not appropriately fit to the experimental data. In fact, the corresponding determination coefficients were low: 0.553 and 0.883 for Cd and Cu, respectively. The calculated APE between the measured and calculated adsorbed masses were high since they are about 28% for Cd and 22% for Cu. Figure 2 confirms this finding since the dashed lines are far from the measured data.

**Table 1.** Kinetic parameters of Cd and Cu removal by the used lignite (C0 = 100 mg L<sup>−</sup>1,D=2gL−1; t = 60 min; pH = 5; T = 20 ± 2 ◦C).


In contrast, the PSO model fits well the experimental data for both metals. The related determination coefficients (>0.89) were much higher than those for the PFO model. Furthermore, the APE for Cd and Cu were determined to only 7% and 12% for Cd and Cu, respectively and therefore were lower than those obtained for the PSO model. The theoretical adsorbed amounts of Cd and Cu at equilibrium (qe,theo), were very close to the experimental ones (Figure 2 and Table 1), with different percentages of about 1% and 3%, respectively. Therefore, under the used experimental conditions, the PSO model is more suitable in fitting the Cd and Cu removal onto lignite. This model suggests that the rate limiting step might be chemical adsorption involving valency forces through sharing or exchange of electrons between these two metals and lignite [46].

The analysis of the adsorption of Cd and Cu onto the used raw lignite through the application of film and intraparticle diffusion models indicated clearly that that the adsorption process proceeds by surface interactions at earlier stages (short times, less than 5 min) and by intraparticle diffusion at later stages (Figure 2). For Cd, the film diffusion coefficient (through boundary layer) is about two times higher than the one of the intraparticle diffusion. This explains the very high adsorption percentage observed after only 1 min (Figure 2). This finding confirms that intraparticle diffusion process controls significantly the rate of Cd adsorption onto the lignite. Similar observations were reported by Pehlivan et al. [32] when studying Pb, Cu, Ni, and Zn removal by Turkish lignites. For Cu, the film and intraparticle diffusion coefficients are quite similar (Table 1) indicating that both processes (boundary layer and intraparticle diffusion) control its removal by the raw lignite.

#### 3.2.3. Effect of Initial Aqueous pH

The effect of the initial aqueous pH on Cd and Cu removal efficiency was performed under the experimental conditions presented in Section 2.3. The experimental results showed that for both metals, the removed amounts increased with the increase of the aqueous pH (Figure 3). As such, for an initial pH of 2, the adsorbed amounts of Cd and Cu were 10.9 and 5.4 mg g<sup>−</sup>1, respectively. At a pH of 5, these quantities reached 26.1 and 18.6 mg g−<sup>1</sup> which are about 2.4 and 3.4 times higher than those registered at an initial pH of 2. This outcome could be explained by the lignite surface charge that is positive at aqueous pH lower than the pHZPC (3.6), which will repulse the two cationic metals. Furthermore, at this pH range, abundant H+ ions in the solution will compete with Cd and Cu over the adsorption sites. However, when the used aqueous pH value is higher than the pHZPC, the lignite surface will carry more negative charges and will consequently favor Cd and Cu adsorption through electrostatic reactions. At this pH range, H+ and other exchangeable cations such as Ca2+, Mg2+, Na+, and K+ could be exchanged with these metals and released in the aqueous solutions. Similar trends were reported by several studies dealing with heavy metals or cationic pollutant adsorption onto raw or modified lignites [32,38] and other adsorbents [35,42,47,48]. For instance, Pehlivan et al. [32] showed that increasing pH form 2 to 6 increased Cu removal efficiency yield by a raw Turkish lignite from about 8% to more than 88%.

#### 3.2.4. Effect of Lignite Dosage

The impact of lignite dose on Cd and Cu removal from aqueous solutions at an initial concentration of 100 mg L<sup>−</sup>1, an initial pH of 5 and 60 min of contact time is given in Figure 4. Their removal yields increase with the increase of the lignite dose. As such, rising the dose from 0.4 to 3 g L−1, Cd and Cu removal efficiencies increased from about 40.5% and 19.8% to 61.9% and 46.7%, respectively. This important increase is linked to the presence of more available adsorption sites that could interact with Cd or Cu when using higher doses. Starting from a used dose of 3.5 g L<sup>−</sup>1, Cd or Cu removal efficiency remains approximately constant due to the saturation of lignite particles. Such high removal efficiencies observed for relatively low lignite doses is a real asset for larger applications at industrial levels.

A similar trend was reported by Binabaj and Ramezanian [28] and Gurses et al. [49] when studying chromium(VI) and methylene blue removal by raw lignites, respectively.

**Figure 3.** Impact of initial aqueous pH on Cd and Cu removal efficiency by lignite (C0 = 100 mg L<sup>−</sup>1,D=2gL−1; t = 60 min; T = 20 ± 2 ◦C).

**Figure 4.** Impact of lignite dose on Cd and Cu removal yields (C0 = 100 mg L<sup>−</sup>1, t = 60 min; pH = 5; T = 20 <sup>±</sup> <sup>2</sup> ◦C).

#### 3.2.5. Competing Effect

Industrial effluents generally contain a complex mixture of organic and non-organic pollutants. The effect of the presence of Pb and Zn on the adsorption of Cd and Cu by the raw lignite was performed according to the experimental conditions described in Section 2.3. The experimental results (Figure 5) showed that for single mode, as reported by Pentari et al. [30], the adsorption efficiency of the studied metals was as follows: Pb > Cd > Cu > Zn. The corresponding removal percentages were estimated at 100%, 74.6%, 71.7% and 44.8%, respectively. Similar observations were also reported by Pehlivan et al. [34] who studied the adsorption of Pb, Cd, Cu, Ni, and Zn on several Turkish lignites. They showed that for the majority of the studied lignites, Pb, Cu and Cd were the most adsorbed metals whereas Ni and Zn were the less adsorbed ones.

**Figure 5.** Cd, Cu, Pb and Zn removal yield in single and multicomponent systems (C0 = 30 mg L−<sup>1</sup> for all metals; D=2gL<sup>−</sup>1; t = 60 min; pH = 5; T = 20 ± 2 ◦C).

In the multicomponent system, all the removal efficiencies of the four heavy metals significantly decreased due to competition. Accordingly, the highest removal decrease was observed for Cd (37.1%) and Zn (29.2%) and the adsorption pattern shifted to: Pb > Cu > Cd > Zn with removal efficiencies of 91.7%, 56.8%, 37.5% and 15.6%, respectively. Consequently, when treating real wastewater containing a mixture of heavy metals, Pb and Cu will be favorably retained by the current lignite compared to other metals. Similar trends were reported by Allen and Brown [50] when investigating the removal of Cd, Cu and Zn by an Irish lignite in a single and multicomponent systems. Their experimental results showed that in single metal sorption mode, the adsorption ability decreases in the order Cd > Cu > Zn. However, in the multicomponent assays, this order changed to: Cu > Cd > Zn. This result is in a contradiction with the one reported by Pentari et al. [30] who pointed out that in multicomponent system containing Pb, Cd, Cu, and Zn, copper adsorption was the most significantly influenced by the presence of the other elements with a total yield reduction of about 5%. This behavior should not be only attributed to metal characteristics including size, electronegativity, availability and hydration energy [15] but also to the physico-chemical properties of the lignite such as the donor atoms abundance (oxygen, nitrogen, sulphur [51]).

#### 3.2.6. Isotherm Adsorption

For the experimental conditions cited in Section 2.3, the data of Cd and Cu adsorption isotherms in comparison with the predicted ones using Freundlich, Langmuir and D-R models are given in Figure 6. Table 2 gives the parameters of these models as well as their fitness to the experimental data.

**Figure 6.** Isotherm experimental and fitted data with Freundlich, Langmuir and Dubinin–Radushkevich (D-R) models for Cd (**a**) and Cu (**b**) removal by lignite (D=2gL<sup>−</sup>1; t = 60 min; pH = 5; T = 20 <sup>±</sup> <sup>2</sup> ◦C; Exp.: Experimental value).


**Table 2.** Adsorption isotherm parameters of Cd and Cu removal by lignite (D = 2 g L<sup>−</sup>1; t = 60 min; pH = 5; T = 20 ± 2 ◦C).

It can be clearly demonstrated from Figure 6 and Table 2 that the three applied models fit very well with the experimental data. Nevertheless, the highest determination coefficients (R2) (0.990 for Cd and for Cu) and the lowest APE (2.3% for Cd and 1.5% for Cu) were observed for D-R model. This model predicts, however, very high adsorption capacities for both Cd and Cu with respective values of 124.8 and 40.5 mg g<sup>−</sup>1, respectively. These values are unrealistic considering the isotherm curves shape (Figure 6) and are mainly imputed to this model's used assumptions especially the one related to the uniformity and homogeneity of the adsorbent's microporous structure [35]. The calculated free energy (E = 1/√2β) was assessed to 10.7 and 13.4 kJ mol−<sup>1</sup> for Cd and Cu, respectively. Both of them are higher than 8 kJ mol−1, indicating that Cd and Cu removal by the used lignite was mainly chemical [49]. It is worth mentioning that compared to Langmuir model, The Freundlich one fits better the experimental data with higher R2 coefficients and lower APE values (Table 2). This outcome indicates that both Cd and Cu adsorption by the used lignite occurs heterogeneously on multilayer surfaces through chemical processes [28].

On the other hand, the Freundlich constant 'n' values were 2.3 and 3.9 for Cd and Cu, respectively. They are in the range of 1–10, which indicates that the adsorption of these two heavy metals by the used lignite is a favorable process. Values in the same range were determined by Pentari et al. [30] when studying Cd and Cu removal by a raw Greek lignite. Moreover, the highest Langmuir's parameter values (obtained for the lowest used initial concentrations) '*RL* <sup>=</sup> <sup>1</sup> <sup>1</sup>+*KL*<sup>∗</sup> *<sup>C</sup>*<sup>0</sup> ' were estimated to 0.41 and 0.24 for Cd and Cu, respectively. They are lower than 1, suggesting that lignite could be considered as a favorable material for Cd and Cu retention from aqueous solutions.

The Langmuir's maximum adsorption capacities of Cd and Cu were determined to 38.0 and 21.4 mg g−1, respectively. A comparison of the used lignite efficiency in removing Cd and Cu with other lignites (Table 3) clearly shows that it can be considered as a promising medium for the removal of heavy metals form wastewater. As a matter of fact, its Cd adsorption capacity was about 5.6, 1.7 and 1.5 times higher than a lignite activated carbon from China [52], a HNO3 treated lignite activated carbon [52], and a raw lignite form Greece [43]. Its Cu adsorption capacity was higher than a Beypazari lignite [51], Ilgin lignite and Beysehir lignite [32]. However, other tested lignites exhibited higher adsorption capacities [27,30,53]. It is important to underline that the observed lignite adsorption capacities of Cd and Cu are also higher than various other agricultural, animal and industrial wastes [54,55]. As compared to natural agricultural wastes, Cd adsorption capacity of the studied lignite was about 7.5 and 2.3 higher than corncob [56] and rice husk [57], respectively. Moreover, Cu removal was 2.8 and 2.1 times more important than Banana peels [58] and grape stalks [59], respectively.


**Table 3.** Comparison of the used lignite removal efficiency of Cd and Cu with other lignites from various regions.

As illustrated in Table 3, Cd or Cu adsorption efficiency by lignites depend not only on their specific surface area (case of the activated lignites [52]) but also on their functional group types and richness and the experimental conditions. The same trend was reported by Puglla et al. [19] who found relatively low adsorption capacities of Cd onto biochars derived from peanut shell (1.038 mg g<sup>−</sup>1), chonta pulp (0.655 mg g−1) and corncob (0.857 mg g<sup>−</sup>1) even if they have interesting textural properties. It is important to underline that the structural and textural properties of the lignite could be improved through specific physical/chemical/thermal modifications [52,60,61]. For instance, Sun et al. [52] found that the chemical/thermal modification of lignite by a HNO3 solution simultaneously increased its surface area (from 158.1 to 185.1 m<sup>2</sup> g<sup>−</sup>1) and its surface carboxylic and lactone functional groups (from 0.55 to 1.26 mmol g<sup>−</sup>1). Consequently, such modification increased Cd removal capacity from 6.8 to 26.6 mg g<sup>−</sup>1.

#### *3.3. Raw and Metal-Loaded Lignite Characterization and Adsorption Mechanism Exploration*

Various techniques were used for the characterization of lignite before and after metal adsorption in order to get a better understanding of the probably involved mechanisms. Morphological properties of the lignite before and after adsorption of metals were investigated through imagery comparison using scanning electron microscopy and energy-dispersive X-ray spectroscopy (Figures 7 and 8).

**Figure 7.** SEM images of: (**a**) raw lignite; (**b**) after adsorption of cadmium; (**c**) after adsorption of copper (magnitude: ×2500).

It appears that the raw lignite (particle size lower than 63 μm) presents an irregular and heterogeneous surface. Moreover, similarly to the majority of natural carbonaceous materials, the used lignite does not present a concrete porosity (Figure 7a), which explain its measured specific surface area. SEM imaging reveals also the presence of crystalline structures on the surface. According to Zhang and Chen [62], raw uncrushed lignites could present some typical cellular plant structures such as fusinites and semifusinites, intergranular and plant tissue pores along with some crystalline impurities identified mainly as calcium carbonates and silicon dioxide as well as other trace elements such as aluminum and iron ions. This was confirmed by EDX analysis where the raw material presents high peak intensities of oxygen, silica and calcium (Figure 8a). After the adsorption process, SEM images of the metal-loaded lignite presented a clearer contrast on its surface for both metals, while the crystalline structure present in the raw material became less apparent (Figure 7b,c). This observation could be due to an adsorption reaction of metals on the lignite surface and at the same time to a possible inverse movement of some minerals from lignite to the liquid phase. The EDX analysis confirmed this hypothesis since the peaks related to oxygen, silica, potassium, and calcium decreased in intensity, while peaks attributed to copper and cadmium appeared, which confirms the adsorption process (Figure 8b,c).

The proximate analysis (Figure 9) indicates that the raw lignite is mainly composed of minerals with an ash content of about 39%. Its moisture, fixed carbon and volatile matter contents are respectively 7%, 21% and 33%. Similar trends were reported by Kanca [63] for a Turkish lignite, where ash content had the major proportion (48%) followed by volatile matter and fixed carbon (25% and 24%, respectively).

**Figure 8.** EDX analysis of: (**a**) raw lignite; (**b**) cadmium-loaded lignite; (**c**) copper-loaded lignite.

**Figure 9.** Proximate analysis of lignite before and after copper and cadmium adsorption.

After metal adsorption, a slight decrease of the fixed carbon and volatile matter contents occurred in favor of an ash content increase by about 3.7% and 5.8% for Cd- and Cu-loaded lignite, respectively. To align these results with SEM/EDS observations, it is possible that the adsorption of cadmium and copper were driven by three mechanisms: (i) adsorption of a small fraction of metal ions onto lignin decayed matrix which constitutes the small specific surface area of the feedstock yet characterized, (ii) ion-exchange reaction where ions such as potassium, calcium then silica at lower percentage were diffused from the solid matrix to the aqueous solution thus vacating the oxygenic functional groups, and (iii) chemical adsorption on free surface functional groups.

In order to confirm this hypothesis, FTIR analyses were performed on both raw and metal-loaded lignite and results are depicted in Figure 10 and Table 4. The lignite presented a heterogeneous surface with the co-presence of acidic and alkaline functional groups namely, hydroxyl (–OH; 3600–3200 cm<sup>−</sup>1), aliphatic (C–H; 3000–2700 cm−1), carbonyl and acetyl esters (C=O and C–O; 1650–1600 cm−<sup>1</sup> and 1180–980 cm−1, respectively), methyl and methylene aromatic (–CH2/–CH3; 1465–1320 cm<sup>−</sup>1) and out-of-plane aromatic groups (C–H; 896–809 cm−1). After metal adsorption, few modifications were noticed for peak positions of some functional groups due to their involvement in the adsorption process (Table 4). For instance, a peak shift of about +8 and +4 cm−<sup>1</sup> was noticed for hydroxyl groups after Cd and Cu adsorption, respectively. Similar observation for carboxylic group where a vibration was spotted by about −4 and −6 cm−<sup>1</sup> as compared to raw lignite for the same exhausted adsorbents, respectively. Moreover, a significant change was observed in the—C=C aromatic structure where a peak appeared for both Cd- and Cu-related specters at 1382 and 1380 cm<sup>−</sup>1, respectively (Table 4). This could be related to the uptake of nitrate ions (entering in the composition of copper and cadmium reagents) by the lignite matrix, causing a slight alteration in its surface functionalities [64].

**Figure 10.** FTIR analysis of lignite before and after Cu (II) and Cd (II) adsorption.

**Table 4.** Peak location for each functional group detected in FTIR spectra for raw and metal-loaded lignite.


On the other hand, the presence of positively charged metals (i.e., Cd2+ and Cu2+) caused a thermodynamic balance between solid and liquid phases. It is possible that a cation release phenomenon occurred leading to the release of protons (H+) or other cations (Na+; K+; Ca2+, and Mg2+) followed by the fixation of bivalent metals on the oxygenic functional groups as follows [19,65]:

$$2\left[\text{A}-\text{C}-\text{O}^{-}-\text{H}^{+}\right] + \text{Me}^{2+} \leftrightarrow \left[\text{A}-\text{C}-\text{O}^{-}-\text{Me}^{2+}\right] + 2\text{H}^{+} \tag{5}$$

$$2\left[\text{A}-\text{O}^{-}-\text{H}^{+}\right] + \text{Me}^{2+} \leftrightarrow \left[\text{A}-\text{O}^{-}-\text{Me}^{2+}\right] + 2\text{H}^{+} \tag{6}$$

$$\left[\text{A}-\text{Na}^+/\text{K}^+/\text{Ca}^{2+}/\text{Mg}^{2+}\right] + \text{Me}^{2+} \leftrightarrow \left[\text{A}-\text{Me}^{2+}\right] + 2\text{Na}^+/2\text{K}^+/\text{Ca}^{2+}/\text{Mg}^{2+} \quad \text{(7)}$$

where A stands for the aromatic structure of lignite and Me2+ is either Cu2+ or Cd2+.

On the basis of FTIR investigation, we can deduce that Cd and Cu removal by lignite is governed not only by its textural properties (especially surface area and microporosity), but also by its richness in functional groups.

#### **4. Conclusions**

The current research study demonstrates that lignite, as a low cost and abundant material, can achieve very rapid and effective removal of cadmium and copper from aqueous effluents under wide experimental conditions. The removal efficiency seems not only dependent on the heavy metal properties but also on the textural and structural characteristics

of the lignite. The heavy metal adsorption process occurs through a combination of several mechanisms, including mainly cation exchange and complexation with various functional groups. Lignite adsorption capacity could be further enhanced through physical, chemical, and thermal treatment methods. However, the cost and the environmental impact of such modifications should be accurately assessed. Moreover, further studies are required in order to assess the efficiency of raw/modified lignite for the removal of other heavy metals and recalcitrant organic pollutants under dynamic conditions.

**Author Contributions:** Conceptualization, S.J. and A.M.; methodology, S.J., A.M. and M.J.; validation, S.J., A.M. and A.A.A.; formal analysis, S.J., A.A.A., M.J., H.H. and A.M.; investigation, S.J. and A.M.; resources, S.J. and A.M.; writing—original draft preparation, S.J. and A.A.A.; writing—review and editing, A.M., H.H. and M.J.; visualization, S.J., A.M., A.A.A., H.H. and M.J.; supervision, S.J. and A.M.; project administration, A.M.; funding acquisition, A.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by The Tunisian Ministry of Higher Education and Scientific Research.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data available on request.

**Acknowledgments:** Authors would like to thank all technicians involved in this work. The Tunisian Ministry of Higher Education and Scientific Research is gratefully acknowledged for financing this research project.

**Conflicts of Interest:** The authors declare no conflict of interest

#### **References**


### *Article* **Advanced Treatment of Real Grey Water by SBR Followed by Ultrafiltration—Performance and Fouling Behavior**

#### **Gabriela Kami ´nska and Anna Marszałek \***

Department of Water and Wastewater Engineering, Silesian University of Technology, Konarskiego 18, 44-100 Gliwice, Poland; gabriela.liszczyk@gmail.com

**\*** Correspondence: anna.marszalek@polsl.pl; Tel.: +48-32-237-21-73

Received: 22 November 2019; Accepted: 3 January 2020; Published: 4 January 2020

**Abstract:** Grey water has been identified as a potential source of water in a number of applications e.g., toilet flushing, laundering in first rinsing, floor cleaning, and irrigation. The major obstacle to the reuse of grey water relates to pathogens, nutrients, and organic matter found in grey water. Therefore, much effort has been put to treat grey water, in order to yield high-quality water deprived of bacteria and with an appropriate value in a wide range of quality parameters (Total Organic Carbon (TOC), nitrate, phosphate, ammonium, pH, and absorbance), similar to the values for tap water. The aim of this study was to treat the real grey water, and turn it into high-quality, safe water. For this purpose, the real grey water was treated by means of a sequential biological reactor (SBR) followed by ultrafiltration. Initially, grey water was treated in a laboratory SBR reactor with a capacity of 3 L, operated in a 24 h cycle. Then, SBR effluent was purified in a cross-flow ultrafiltration setup. Treatment efficiency in SBR and ultrafiltration was assessed using extended physicochemical and microbiological analyses (pH, conductivity, color, absorbance, Chemical Oxygen Demand (COD), Biological Oxygen Demand (BOD5), nitrate, phosphate, ammonium, total nitrogen, phenol index, nonionic and anionic surfactants, TOC, *Escherichia coli*, and enterococci). Additionally, ultrafiltration was evaluated in terms of fouling behavior for three polymer membranes with different MWCO (molecular weight cut-off). The values of quality parameters (pH, conductivity, COD, BOD5, TOC, N-NH4 <sup>+</sup>, N-NO3 <sup>−</sup>, Ntot, and P-PO4 <sup>3</sup>−) measured in SBR effluent did not exceed permissible values for wastewater discharged to soil and water. Ultrafiltration provided the high-quality water with very low values of COD (5.8–18.1 mg/L), TOC (0.47–2.19 mg/L), absorbanceUV254 (0.015–0.048 1/cm), color (10–29 mgPt/L) and concentration of nitrate (0.18–0.56 mg/L), phosphate (0.9–2.1 mg/L), ammonium (0.03–0.11 mg/L), and total nitrogen (3.3–4.7 mg/L) as well as lack of *E. coli* and enterococci. Membrane structural and surface properties did not affect the treatment efficiency, but did influence the fouling behavior.

**Keywords:** grey water; SBR; ultrafiltration; fouling; zeta potential

#### **1. Introduction**

Water is emerging as one of the single most important resources of Planet Earth for the prosperity of the economy and human life. However, freshwater resources have been increasingly polluted and depleted globally. The constant increase in water usage and climate change—such as altered weather-patterns (including droughts or floods), deforestation, and increased pollution—are the main driving forces for the rising global scarcity of water. It is experienced especially by European countries, due to an improvement of living standards and economic development in recent years. Along with economic development, the total water use in Europe has been sharply increasing in the last decades [1]. Grey water recycling is the solution to the growing challenges associated with water shortages [2]. Quantitatively, grey water represents around 65% of the total volume of domestic sewage, making it the

largest stream for water reuse [3]. Grey water contains nitrate, phosphate, organic matter, surfactants, pharmaceuticals, oils, and pathogens [4]. Therefore, the treatment of grey water before its reuse is necessary. Purified grey water could be reused for cleaning, car washing, concrete production, and irrigation [5,6]. The guidelines of the World Health Organization (WHO) point to four criteria for the reuse of grey water: hygiene, aesthetics, environmental tolerance, and economic feasibility. WHO recommends biological treatment and ultrafiltration in order to obtain high-quality water. However, the requirements that define "high water quality" are too sweeping and do not show the physical and chemical specifications. It is caused by a number of possible applications of reclaimed grey water that require various different quality standards. In 2006, WHO set microbiological criteria that reclaimed grey water should meet for its reuse for restricted and non-restricted agricultural irrigation. [7,8].

Numerous approaches, including low and high pressure-driven membrane techniques [9–11], coagulation [12,13], biological processes, and membrane bioreactors [3], have been proposed for the treatment of grey water. These technologies vary in both complexity and performance. For example, Li et al. recommended the employment of an aerobic treatment coupled with membrane filtration or an aerobic treatment with sand filtration, with disinfection as the last step for both systems [8]. Similarly, Ding et al. suggested applying a system that combines membrane filtration with biological treatment in a gravity-driven membrane filtration system. In this technique, the fouling layer attached to the membrane stabilizes a flux and improves the treatment effects due to the microbial activity of biofilm [14]. From these studies, it is clear that grey water requires both biological and physical treatment to meet non-potable reuse standards. Aerobic biological processes are effective to remove organics from grey water. Physical processes, particularly membrane filtration, are recommended for polishing effluents from the biological treatment step [9,15]. Ultrafiltration (UF) retains bacteria, suspension, colloids, natural organic matter, and partial micropollutants such as pharmaceuticals and personal care products. However, the efficiency of ultrafiltration in the removal of individual pollutants depends on membrane type and its properties such as molecular weight cut-off (MWCO). Membrane properties such as contact angle and zeta potential play an important role in the fouling behavior and hydraulic performance of the system [16]. Membrane fouling results in a permeability loss due to the increase in hydraulic resistances in the filtration system, especially in the case of porous polymer membranes. There are many studies reported in the literature that aimed at mitigating the fouling by selecting the most optimal operational parameters for ultrafiltration such as transmembrane pressure, velocity, temperature, or developing a cleaning method. [17–19]. On the other hand, they do not consider the effect of structure and surface properties of membrane i.e., MWCO and contact angle on fouling behavior. Another issue is that most of these studies were conducted for artificial grey water. While artificial grey water provides useful information for model development, it does not reflect real conditions, and the composition affects the process significantly.

The main novelty of this work shows that high-quality water can be obtained from grey water. For a need of this study, we established that high-quality water is water: (1) that fulfills criteria included in the Polish Regulation from the Minister of the Environment on the conditions to be met when discharging wastewater into water or soil and (2) with the basic quality parameters (smell, color, turbidity, TOC, ammonium, nitrate, chloride, conductivity, hardness, *E. coli*, and enterococci) like the standards for tap water. In order to gain this purpose, real grey water was treated using a sequential biological reactor (SBR) followed by ultrafiltration. Treatment efficiency was evaluated for SBR and ultrafiltration using extended physicochemical and microbiological analyses. Additionally, ultrafiltration was studied in terms of fouling behavior for three polymer membranes with different MWCO. In that context, the novelty of this work is also a determination of the effect of membrane type on the treatment of SBR effluent and fouling behavior.

#### **2. Methodology**

#### *2.1. Grey Water Characteristic*

Grey water was collected from a single-family household located in Silesia in Poland. This house is equipped with an installation to collect grey water from the shower, kitchen sink, and washing machine. The characteristic of the grey water is given in Table 1. Samples were taken directly from a storage tank and transported to the lab facilities, stored at 4 ◦C and analyzed within 48 h according to the methodology presented in Section 2.6.


**Table 1.** Characteristic of grey water, as taken.

#### *2.2. Biological Treatment in SBR*

The SBR technology used provides for the dosing of wastewater in alternately used anaerobic reactor oxygen conditions. This guarantees a high degree of removal of carbon, nitrogen, and phosphorus compounds. The operation of the reactor is simple, and during operation, modifications of individual separate phases of the technological cycles are possible. The biological process was carried out under laboratory conditions, using activated sludge taken from the municipal sewage treatment plant in Gliwice, Poland. The treatment of grey water was carried out in a laboratory SBR with a capacity of 3 L. During the treatment experiment, the excessive activated sludge was periodically removed from the SBR in order to keep its concentration at the level of 3.0 g/L. The solid retention time (SRT) was 20 days. The load of the sludge with the contaminants was equal to 0.1 g COD/gDMd and the concentration of oxygen was at the level of 3 mg/L. The system was operated as the sequential biological reactor in one cycle per day. The length of particular operation stages was as follows: filling and mixing phase for 2 h, aeration phase for 21 h, and sedimentation and SBR effluent removal for 1 h. The aeration system consisted of aquarium cubes placed on the bottom of the tank with connected aeration pumps of the Tetratec APS 300 type (Tetra, Melle, Germany). In order to thoroughly mix the contents of the chamber, it was mixed by means of a magnetic stirrer with the possibility of speed regulation in the range from 50 to 1000 revolutions/min. The chamber was filled and emptied using Heidolph peristaltic pumps (Heidolph, Schwabach, Germany). The electronic weekly programmer controlled the start and end of the relevant SBR work phase.

#### *2.3. SBR E*ffl*uent Treatemnt by Ultrafilltration–Ultrafiltration Run*

Ultrafiltration is especially dedicated to drinking water production because colloids, particulates, macromolecules, and pathogens are removed in this process. It is also a suitable process to purify effluent after biological treatment as this effluent contains bacteria, particles of organic matter, and

suspended solids. Ultrafiltration was carried out in the lab, using a scale cross-flow configuration equipped with a plate-and-frame membrane module SEPA CF-NP (GE Osmonics, Minnetonka, MN, USA) as seen in Figure 1. Three ultrafiltration membranes were used in separate processes. Membrane properties are presented in Table 2. Before each experiment, the clean water flux was determined using ultrapure water. The process was operated at a constant pressure of 5 bar and a constant temperature of 22 ± 1 ◦C and a constant velocity of 1 m/s with continuous dosing of SBR effluent to the feed tank. Each filtration run consisted of three cycles including 60 min of filtration followed by forward flushing with ultrapure water during 60 s. The volume of permeate was monitored in order to determine the permeability from the following equation:

$$L\_p = \frac{V}{\Delta p \cdot A \cdot t} \tag{1}$$

where: *Lp* is permeability (L·m−2·h−1·bar<sup>−</sup>1) in short (LMHB), *<sup>V</sup>* is permeate volume (L), *<sup>A</sup>* is membrane surface area (m2), *t* is permeate time collection (h), and Δ*p* is transmembrane pressure (bar).


\* own measurements with an electronic micrometer.


**Figure 1.** Schematic diagram of the cross-flow filtration set-up.

#### *2.4. Membrane Fouling Characterization*

In this study, hydraulic resistances were used to characterize the fouling behavior of ultrafiltration membranes treating SBR effluent. Hydraulic resistances of membrane and fouling layer were calculated using the resistance in the series model and Darcy's law using the correlations as shown below [20].

$$k\_{tot} = k\_{\text{ur}} + k\_f \tag{2}$$

$$k\_f = k\_{irr} + k\_{rev} \tag{3}$$

$$J = \frac{\Delta p}{\mu \cdot k} \tag{4}$$

where: *k* is hydraulic resistance, where subscripts *m*, *f*, *irr*, *rev*, and *tot* are related to membrane, fouling, hydraulically irreversible fouling, hydraulically reversible fouling, and total (m−1), respectively, *J* is the flux (m3·m−2·s<sup>−</sup>1), <sup>Δ</sup>*<sup>p</sup>* is the transmembrane pressure (kg·s−2·m<sup>−</sup>1), and <sup>μ</sup> is the dynamic viscosity of water at given temperature (kg·m−1·s−1). Membrane resistance (*km*) was measured for the clean membrane with ultrapure water prior to feed water filtration. Hydraulically irreversible fouling was determined from the flux after forward flushing, while hydraulically reversible fouling was determined from the difference in fouling and irreversible resistances.

#### *2.5. Membrane Characterization*

Zeta potential of membranes was determined by the electrokinetic analyzer SurPASS ™ 3 (Anton Paar, Graz, Austria). Measurements of the contact angle were performed using the goniometer PG-1 (Fibro System AB, Hägersten, Sweden) and the sessile drop method was applied. By syringe on top, a drop of ultrapure water was put on the dried membrane surface. Through an enlarged projection of the water drop on the gauge, the value of the contact angle was measured. For every type of membrane, 10 samples were measured and the average value was calculated.

#### *2.6. Quality Analysis and Microbiological Assessment*

The treatment efficiency in SBR and ultrafiltration was evaluated by the monitoring of the typical quality parameters (color, turbidity, COD, BOD5, TOC, phenolic index, absorbance of UV254, nitrate (N-NO3 <sup>−</sup>), phosphate (P-PO4 <sup>3</sup>−), ammonium (N-NH4 <sup>+</sup>), total nitrogen (Ntot), conductivity, pH, and anionic and non-ionic surfactants). Color and turbidity measurements were performed with a UV-Vis Spectroquant®Pharo 300 (Merck, Kenilworth, NJ, USA). Phenolic index, COD, nitrate, phosphate, ammonium, total nitrogen, and anionic and non-ionic surfactant concentrations were determined spectrophotometrically with Merck test kits (codes of the Merck test kits are given in Table S1). The absorbance was measured at 254 nm, using a UV-visible light (UV-Vis) Cecil 1000 (Analytik Jena AG company, Jena, Germany). TOC was measured using a TOC-L series analyzer (Shimadzu, Kioto, Prefektura Kioto, Japan). pH and conductivity were monitored by multifunctional analyzer CX-461 (Elmetron, Zabrze, Poland). The BOD5 was determined by respirometric measurement using the OXI Top System WTW set (Xylem Analytics, Weilheim Germany). Microbiological analysis including *E. coli* and enterococci was conducted by an external accredited lab according to ISO methods PN-EN ISO 9308-1:2014-12/PN-EN ISO 9308-1/A1:2017; PN-EN ISO 7899-2:2004.

#### **3. Results and Discussion**

#### *3.1. Grey Water Treatment in SBR*

#### 3.1.1. Reduction of COD, BOD5, and TOC

The biological sewage treatment aimed to reduce biodegradable organic matter and remove the biogenic substances i.e., nitrogen and phosphorus. In the preliminary stage, the activated sludge was taken from the municipal wastewater plant, then it was adapted to a new influent (grey water). The results discussed in Figures 2–4 show the analysis of individual parameters after the adaptation process. Figure 2 shows the SBR system performance in relation to the concentration of organic compounds in purified grey water. Detailed results of physicochemical analyses are presented in Table S2 (in Supplementary file).

**Figure 2.** The performance of the sequential biological reactor (SBR) system with respect to the changes in the concentration of the organic compounds.

It was shown that the degree of pollution removal was high, and the values of the quality parameters in SBR effluent did not exceed the permissible values included in the currently in force in Poland, Regulation of the Minister of the Environment on the conditions to be met when discharging

wastewater into water or soil [21]. Over the whole experimental period (22 days), the SBR system reduced COD by around 92%, from an average value of 543 mg/L (influent) to 44 mg/L (effluent) (Figure 2a). Fountoulakis et al. obtained a similar result in their studies [22]. More specifically, the removal of COD was approximately 87%. Such, a high removal was also obtained in studies on grey water purification in a membrane bioreactor operating in a flow-through system [3], where the COD reduction degree was 88%.

It was found that the easily biodegradable compounds expressed by BOD5 were removed in 99% of cases, and the average concentration in the effluent was 4 mg/L (Figure 2b). Similarly, a high reduction was recorded for TOC and absorbance, i.e., 91% and 76%, respectively (Figure 2c,d). Anionic and nonionic surfactants were reduced by 97% and 100%, respectively (as seen in Table S1 in the Supplementary File). In the study [22], the removal efficiency of the anionic surfactants was about 80% in the submerged membrane bioreactor. This is due to the fact that non-ionic and anionic surfactants are readily biodegradable under aerobic conditions [23–25].

#### 3.1.2. Removal of Biogenic Compounds

As seen in Figure 3, the concentration of nutrients in SBR effluent was low and did not exceed permissible values laid out in the Polish regulations [21].

**Figure 3.** The performance of the SBR system with respect to the changes in the concentration of the nutrients. GW—grey water, EF—effluent.

Raw grey water was characterized by a minimum content of ammonium and nitrate nitrogen at the level of 0.2 mg/L and 0.8 mg/L, respectively. After biological treatment with the SBR system, the concentration of nitrate and nitrogen increased to an average value of 7.6 mg/L. Differences in the nitrate concentrations in the influent and effluent indicate that a large part of nitrogen in grey water was organically bound [16]. The average value of total Kjeldahl nitrogen in the influent was 10.6 mg/L. The removal degree of total nitrogen was at the average level of 20% and the concentration in the effluent was 8 mg/L. It should be emphasized that despite the lack of total denitrification, the concentrations of individual forms of nitrogen in SBR effluent did not exceed the maximum permissible values given that are currently in force in Poland, (Regulations of the Minister of the Environment) on the conditions to be met when discharging wastewater into water or soil [21]. At the initial stage of the grey water purification process, there was a problem with phosphorus removal. Its value exceeded several times the permissible value. It can be explained by the variable physicochemical nature of the SBR influent. The effective dephosphatation takes place when the influent contains an easily biodegradable COD fraction and the COD/BOD5 ratio is 2 [26,27]. The second factor intensifying a release of phosphate under anaerobic conditions is the constant delivery of easily biodegradable organic compounds, e.g., volatile fatty acids and their salts. Their occurrence in grey water is likely to be changeable. Since grey water was collected in a holding tank over a longer time, it could undergo an acid fermentation during which volatile fatty acids may be formed. It was found that in the second week of the process, the degree of P-PO4 <sup>3</sup><sup>−</sup> removal increased along with the increase of COD concentration of the inflowing grey water. In the following weeks of operation of the SBR reactor, the phosphate phosphorus concentration ranged from 1.1 to 2.6 mgP-PO4 <sup>3</sup>−/L. Despite the increase in efficiency of phosphorus removal, its permissible concentration specified in the Regulation of the Minister of the Environment [21] (Ptot = 2 mg/L) was still exceeded. In order to improve the effectiveness of phosphate removal, the cycle of operation of SBR should be modified by means of changes in the duration of aerobic–anaerobic phases [28].

#### 3.1.3. Removal of Color and Turbidity

Figures 4 and 5 show the treatment efficiency of the SBR system with respect to the changes in the color and the turbidity. From these results, it is clear that over the whole experiment, the efficiency of the SBR system in terms of color and turbidity was at a constant high level. During 22 days of continuous SBR system operation, the color of grey water decreased from an average value of 201 mgPt/L to 44 mgPt/L, which corresponds to an average removal of 78%. Meanwhile, the average grey water turbidity decreased from 28 FTU to 4 FTU corresponds to an average removal of 84%.

**Figure 4.** The removal performance of the SBR system with respect to the changes in the color and the turbidity.

**Figure 5.** Changes in the color and the turbidity of grey water under treatment with SBR followed by ultrafiltration.

#### *3.2. SBR E*ffl*uent Treatment by Ultrafiltration*

The treatment efficiency of SBR effluent in an ultrafiltration unit with different membranes is presented in Figure 6 and Table S3 (Supplementary file). As expected, for ultrafiltration, removal of organics expressed by the color, absorbance, COD, and TOC was very high. More specifically, the color, absorbanceUV254, COD, and TOC were reduced maximally by 73%, 91%, 84%, and 91%, respectively. Slightly lower elimination was observed for inorganic quality parameters. Conductivity, nitrate, phosphate, ammonium, and total nitrogen were reduced by 61%, 64%, 79%, 85%, and 59%. It is important to emphasize that negative charge of membranes at pH 7–8 (as seen in Figure 8) played an important role in the reduction of conductivity, phosphate, and nitrate ions in SBR effluent. The

literature describes that negatively charged ultrafiltration membranes reject phosphate ions by 87%, as an effect of electrostatic repulsion [29]. Another important influencing factor on the rejection of ions and organics in the ultrafiltration unit treating wastewater/surface water can be hydrophobic/hydrophilic interactions between organics and ions in feed water [30]. For example, Shang et al. reported that high phosphate removal was attributed to the adsorption of phosphate on biopolymers in effluent and removal with these biopolymers [31].

**Figure 6.** Reduction values of organic (**a**) and inorganic (**b**) parameters during the ultrafiltration of SBR effluent. Each bar corresponds to the filtration cycle 1–3.

Surprisingly, membrane DSGM with the lower value of MWCO provided permeate quality values very similar to the values for BN and V3 membranes with higher MWCO. This brings an important finding that the treatment efficiency was more affected by operational conditions of cross-flow ultrafiltration than by membrane properties. When good mixing and turbulent flow along the membrane surface are guaranteed, as it is in cross-flow, other factors affecting permeability and selectivity are less important [32]. Feed nature could also play an important role. From the quality parameters of SBR effluent, we can assume that SBR effluent did not contain a significant portion of low molecular weight organics that needed to be removed. Another reason could be the fouling layer that supported the retention of pollutants by BN and V3 membranes. Jerman et al. found that the cake or gel layer acts as an additional membrane in ultrafiltration and improves retention by up to 40% [33].

It was also found that treatment efficiency increased within cycles 1–3 for all membranes (Figure 6). It can be related to the formation of the cake layer on the membrane surface of BN and V3. This corresponds well with the permeability loss observed for these membranes. Many authors suggest that the cake layer decreases membrane permeability by the reduction of effective pore size and improves the retention of organics and ions [34–36]. However, in the case of the DSGM membrane, there was no fouling so the cake layer was probably not created. The reason for increasing retention over time for DSGM can be related to the concentration effect of feed components within cycles

1–3. In the literature, the influence of the feed concentration on both ions and organics removal can be found [37,38]. Muthumareeswaran et al. have found that a change of the concentration of feed components in the multicomponent system affects their retention due to the ionic interaction between the feed components (Columbic interaction), molar volume, and interaction between ions and membrane surface charge [39].

Importantly, physical, chemical, and microbiological specifications (smell, color, turbidity, TOC, ammonium, nitrate, chloride, conductivity, hardness, *E. coli*, and enterococci) of the permeate correspond well with typical tap water values of basic quality parameters [40].

#### *3.3. Microbiological Quality of Grey Water*

Pathogens, such as *Escherichia coli* and Enterococci, have been identified in grey water (Table 3). The concentration of these bacteria was above 100 CFU per 100 mL. During the biological purification process in the SBR system, the presence of *Escherichia coli* did not change, while Enterococci decreased to an average of 3 CFU per 100 mL of sample. Then, ultrafiltration for each UF membranes resulted in the complete removal of the determined pathogens [41]. High pathogen removal from the *E. coli* group was also noted in the SMBR system from Khalid Bani-Melhem et al. [3].

**Table 3.** Microbiological quality of grey water.


#### *3.4. Membrane Permeability and Fouling Behavior in Ultrafiltration*

Hydraulic performance of membranes was evaluated by permeability loss as a function of time. As seen in Figure 7, the permeability decreased gradually for BN and V3 membranes, while for DSGM, permeability was constant along the ultrafiltration. In other words, fouling was not observed for the DSGM membrane. This finding is very important in the context of membrane lifetime and reducing operational cost. Similarly, Acero et al. reported that UF membranes (with higher MWCO) revealed higher fouling than NF membranes (with lower MWCO) [42]. It can be explained by different surface properties of given membranes, such as hydrophilicity/hydrophobicity and surface charge (zeta potential curve and isoelectric point). The contact angle and isoelectric points of clean and fouled membranes are listed in Table 4. Zeta potential curves for clean and fouled membranes are presented in Figure 8. Owing to the lowest contact angle and strong negative surface charge, the DSGM membrane had the best antifouling properties. On the contrary, the surface of BN and V3 membranes was much more hydrophobic, and thus more prone to adsorb the feed components. Some authors reported a high importance of membrane hydrophilicity and negative charge for fouling mitigation in ultrafiltration [43–45] treating water or wastewater.

In order to investigate this further, the hydraulically reversible and hydraulically irreversible resistances were calculated. As seen in Figure 9, the total fouling (sum of reversible and irreversible resistances) was higher for the V3 membrane. However, considering the proportion of reversible and irreversible resistances for these membranes, it is clear that fouling behavior for these membranes exhibited the opposite trend. More specifically, a significant proportion of the increase in resistance for the V3 membrane was hydraulically reversible, while hydraulically irreversible for the BN membrane. Reversibility/irreversibility of fouling depends on the surface properties of membranes. It is in good correspondence with changes of course of zeta potential curves and shift in isoelectric point for clean and fouled membranes. Membrane V3 with the lower negative charge was more sensitive to fouling overall than the BN membrane. However, for the BN membrane, we observed the higher shift in isoelectric point (from 3.01 to 3.55) than for V3 (from 5.1 to 5.25) indicating a persistent change in BN membrane properties caused by irreversible fouling.

**Figure 7.** Permeability loss as a function of time of UF for different types of membranes. Ultrafiltration with DSGM membrane was performed without forward flushing due to constant permeability.



**Figure 8.** Zeta potential vs pH for clean and fouled membranes.

**Figure 9.** Irreversible and reversible fouling resistances of ultrafiltration for all membranes. Each bar corresponds to the filtration cycle 1–3. *kf irr* and *kf rev* are irreversible and reversible resistances, respectively.

#### **4. Conclusions**

This study proved a great performance of SBR followed by ultrafiltration to obtain high-quality water for non-potable purposes from real grey water. Purified grey water fulfilled criteria for wastewater discharged into water or soil as well as physical, chemical, and microbiological requirements for tap water. It is a good starting point to find new reuse application of reclaimed grey water.

The following conclusions can be reached from the experimental results.


**Supplementary Materials:** The following is available online at http://www.mdpi.com/2073-4441/12/1/154/s1, Table S1: Codes of the Merck test kits, Table S2: Values of quality parameters and reduction degree for SBR system, Table S3: Values of quality parameters measured in SBR effluent and permeate samples.

**Author Contributions:** Conceptualization G.K. and A.M. methodology, G.K. and A.M.; investigation, G.K. and A.M.; writing—original draft preparation, G.K. and A.M., Writing—review and editing, G.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Polish Ministry of Science and Higher Education.

**Conflicts of Interest:** The author declares no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**

1. Use of Freshwater Resources. Available online: https://www.eea.europa.eu/data-and-maps/indicators/useof-freshwater-resources-2/assessment-3 (accessed on 23 December 2019).


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*
