Adsorption of Pharmaceutical Contaminants from Aqueous Solutions Using N,O-Carboxymethyl Chitosan/Polyethylene Oxide (PEO) Electrospun Nanofibers

Residues of pharmaceutical and direct metabolites discharged into the aquatic environment have become a challenge for wastewater treatment facilities due to their increase in concentration and their different physicochemical properties. These emerging contaminants are daily detected in surface water and wastewater discharged by municipalities. To remediate the contaminated water, various methods are currently used including primary, secondary, and tertiary advanced treatments. However, some economic and environmental limitations have forced the scientific community to develop alternative disin-fection processes to purify wastewater. As such, the adsorption strategy represents a “green” low-cost and effective solution to remove pollutants from water. In this study, a nanomaterial made of N,O-carboxymethyl chitosan (N,O-CMCS) was prepared using chitosan (CS) and monochloroacetic acid under various conditions. N,O-CMCS electrospun was synthetized with the copolymer polyethylene oxide (PEO) to create nanofiber membranes showing a better specificity toward diversified contaminants depending on the pH of medium. The developed adsorbent was used to remove fluoxetine (FLX) from aqueous solutions. The new nanomaterial was characterised using FTIR, NMR, and SEM techniques. Sorption batch tests were carried out using high-performance liquid chromatography and ultraviolet diode array detector (HPLC-UV DAD) under controlled pH experimental conditions to determine the contaminant removal capacity of the nanomaterial. The prom-How ising adsorption results obtained with N,O-CMCS/PEO nanofibers are among the best ones obtained so far in comparison to other commercial and synthetized adsorbents tested for FLX’s adsorption. Kinetic experiments were also performed to investigate effects of contact times on the FLX adsorption. Experimental results were fitted to both common kinetic models pseudo-first and second order. The latter kinetic model described the best the sorption on surface. It revealed a possible chemisorption mechanism with electrostatic bounding for N,O-CMCS/PEO nanofibers.


Introduction
The contamination of natural resources such as water remains one of the most challenging issues of modern-day society. Worldwide water contamination by chemicals is especially of concern [1]. Even though most of these contaminants are detected at very low concentrations (ng/L to μg/L), many molecules (and/or their treatment by-products) possess biological activity. They are daily discharged into water bodies and may cause severe environmental issues (e.g. detrimental effects to our natural water resources and aquatic organisms) if they are not properly removed [2] [3]. Among listed contaminants, pharmaceutical products are now recognized as primary polluters retrieved in marine environments and ecosystems mainly after human or animal excretion [4] [5] [6]. These pollutants are represented by a wide array of substances that include both non-prescription and prescription drugs [7]. They are part of different classes (e.g. analgesics, antibiotics, β-blockers, lipid regulators, antidepressants, contraceptives, synthetic and natural hormones) [8] [9]. In addition, industrial by-products, as well as some hospital effluents have been reported in literature as major contamination sources since they are released at higher concentrations in municipal effluents [10]. Unfortunately, limited removal rates are frequently observed in wastewater treatment plants (WWTPs) for pharmaceuticals residues [11] [12].
Since traditional water and wastewater treatment systems are facing difficulties to provide an effective barrier against recalcitrant compounds, advanced treatment strategies need to be developed [13]. As such, membrane filtration, advanced oxidation processes (AOPs), and UV irradiation are examples of processes that have the potential to improve the water remediation [14]. Despite their effectiveness in removing pharmaceuticals residues, some of these may not be suitable due to "high-unit-volume" treatment costs. Additional efforts on the extraction methodologies are therefore required, especially with emerging nano-technologies [13]. In this way, adsorption of contaminants is now recognized as a promising and efficient separation technique for wastewater treatment [15].
The adsorption phenomenon on surface's materials can be described using different theoretical developed models of varied complexity. Thus, the uptake rate can be determined with pseudo-first-order (PFO) and pseudo-second-order (PSO) models, while mechanisms of sorption can be elucidated using well-known and recognized isotherms models (e.g. Langmuir, Freundlich, etc.) [16].
Among existing adsorbent materials, zeolites [17] and activated carbon (AC) [18] [19] [20] were largely studied for treating wastewater due to their high performance. However, a longer recycling time and expensive regeneration procedures have made them difficult to use on a regular basis. To overcome such drawbacks, a considerable work was spent in developing effective and environmentally friendly adsorbents based on available low-cost natural polymeric materials [10]. Accordingly, different biomaterials were found to be effective for the elimination of contaminants from aqueous effluent.
In the last decade, biomaterials have received a lot of attention because of their biodegradable and non-toxic composition [21]. The naturally abundant polysaccharide chitosan (CS) produced from a deacetylation of chitin, has been used in several adsorption applications [15]. However, its high viscosity, crystallinity, as well as its poor mechanical strength, limited solubility and instability in acidic medium have restricted its usefulness as adsorbent [22] [23] [24]. Hence, chemical modifications of CS are performed to improve some properties (e.g. solubility, antimicrobial behavior, and ability to interfere with other compounds) [25] [26]. A chemical modification may also be attempted in order to gain more specificity and a sorption versatility toward contaminants, depending on the grafted chemical functionalities on the surface of CS [27].
One well-known modified CS derivative is N,O-carboxymethyl chitosan (N,O-CMCS) [27] [28]. This modified biopolymer is synthetized following an alkalization and a nucleophilic substitution (SN2) reaction using monochloroacetic acid (ClCH 2 COOH) [29] [30]. The medium temperature and reagents of the reaction (including their stoichiometry) influence both the substitution of the CMCS (N-and/or O-) and its ratio of substitution [31] [32]. Aside from its high viscosity and hydrodynamic volume, CMCS has unique chemical, physical and biological properties [33]. This modified biopolymer enables low toxicity, biodegradability, biocompatibility and high ability to form films, fibers, and hydrogels [34]. Some chemical modifications achieved on CS's surface were exploited in order to increase its water solubility and broaden its range of applications [35] [36] [37]. Other carboxymethyl derivatives having different properties were reported elsewhere (O-CMCS, N,O-CMCS, N-CMCS and N-succinyl chitosan (NSC)) [38].
Synthetized nanomaterials are of great interest for a variety of applications due to their sought characteristics, such as high specific surface area and porosity [26]. In this way, electrospinning has gained popularity because of its ability to produce polymer nanofibers with diameters varying from several micrometers to tens of nanometers [39] [40]. During electrospinning, a high voltage is applied to generate an electrically charged jet of a polymer previously dissolved in solution. The elongated jet is then collected on a metallic surface (target electrode) upon the evaporation of the used solvent producing nanofibers as a nonwoven mat. The formation of electrospun nanofibers from polymer solutions has been extensively studied in terms of voltage, tip-to-collector distance, polymer solution extrusion rate, and polymer solution properties [41].
Until now, a wide range of harmful environmental contaminants such as heavy metals, dye materials, and a few pharmaceuticals residues have been adsorbed by CS nanofiber membranes [42] [43] [44]. Therefore, the use of mem-

Preparation of N,O-CMCS
An adapted methodology was used for the synthesis of N,O-CMCS by alkalization followed by carboxylation [45]. CS (2.00 g) was mixed in 20 mL of a sodium hydroxide (20%) solution and stirred at room temperature (RT) for 12 h. The resulting CS of higher degree of deacetylation was separated by filtration. Next, 15 mL of isopropanol was added to the previously treated CS. Then, monochloroacetic acid (1.43 g) dissolved in isopropanol (10 mL) was slowly added dropwise into the treated CS isopropyl alcohol solution. The mixture was agitated for an additional 30 min. The flask was then put into a heated oil bath and stirred for 4 h. Then, the content was poured into a 50 mL beaker. Under continuous stirring condition, acetic acid (50%) was slowly added until the pH value reached 9. The product was in its salt-based form (N,O-CMCS-Na) due to the alkaline reaction medium. Finally, the reaction mixture was filtered, and the solid product was rinsed 3 times with a 200 mL ethanol solution, then once with absolute ethanol. The resulting solid was subsequently dried in an oven (50˚C) for three days. After, 1 g of N,O-CMCS-Na was suspended in 80% ethyl alcohol aqueous solution (100 mL), and then 10 mL of hydrochloric acid (37%) was added dropwise and stirred for 30 min to get the neutralized form. The resulting solid N,O-CMCS was filtered and rinsed using ethanol (70% -90%) prior being dried out under vacuum overnight at RT [20]. To determine the zeta potential (ZP), a 1 M (10 mL) solution of N,O-CMCS was dissolved in distilled water for 1 h with moderate shaking at room temperature. NaOH (0.01 M) was used to neutralize the pH of the solution. The zeta potential in solution was determined using a ZetasizerNano (Malvern Instruments Ltd., model ZEN 3600).  Table 1.

Electrospinning, Preparation of N,O-CMCS and PEO Solutions
All electrospinning solutions were magnetically stirred for 2 h, then transferred in an ultrasonic bath for 15 min to remove air bubbles. Finally, the mixture was rested for 3 h before being used in the electrospinning setup. The N,O-CMCS/PEO solution was poured into an electrospinning plastic syringe or kept in a refrigerator at 4˚C. The voltage used ranged between 10 and 15 kV. Various electrospinning conditions were attempted to obtain an optimal jet as well (stable with minimal drop projections) and beadless nanofibers mats by varying the flow rate, distance, and electrical current. Electrospinning was carried out in a tailor-made enclosure, at RT and relative humidity ranging between 30% -50%. The nanofibrous mat was removed from the frame at the end of the electrospinning process and dried for 24 h in an oven at 80˚C.

Characterization of Nanofibers
The degree of substitution (DS) of the sample was calculated using a potentiometric titration method [46].  [46]. The cumulative DS of the sample was determined using Equations (1) and (2): where M CMCS is the mass of CMCS (g), V NaOH and C NaOH are representing respectively the volume and molar concentration (M) of NaOH, and finally 161 and 58 are associated to the molecular weights of the chitosan (glucosamine) and the CMCS group [31]. Morphologies of both nanofibers CS/PEO and N,O-CMCS/PEO were assessed using a scanning electron microscope (SEM) (Hitachi SU1510 Scanning Electron Microscope). The average diameter of electrospun nanofibers was estimated using SEM images and Image J software. In order to obtain an average value, the diameter of 50 nanofibers collected on three separate images was calculated (e.g. total of 150 nanofibers per sample).
Some infrared (FT-IR) spectra analyses of the electrospun nanofiber mats were performed using a Fourier transform infrared spectrometer (Nicolet iS10, Thermo Scientific) in the 400 to 4000 cm −1 wavenumbers range. In addition, 1 H-NMR spectra of CS and N,O-CMCS in D 2 O/DCl were recorded using a 200 MHz NMR spectrometer (Varian).

Batch Adsorption Procedure
The adsorption of FLX by the electrospun nanofibrous mat was carried out in batch mode. A 25 mg sample of the membrane was inserted into one Erlenmeyer flask containing 50 mL of an FLX solution (50 ppm) and 5% of methanol. The flask was then agitated on an orbital shaker (ORBIT Environ-shaker, Lab-Line) at RT. Before insertion of the membrane into the solution, an aliquot of 500 µL was sampled to determine the initial FLX concentration. Batch tests were con-Journal of Materials Science and Chemical Engineering ducted over a period of 150 min to ensure that equilibrium was achieved prior to the collection of a 500 µL aliquot. The adsorption capacity at time t (Q t ) was determined using the following Equation (3): where C e is the concentration of the contaminant (ppm) at time t (min), C 0 is the initial concentration of contaminants (ppm), V is the volume of the solution (L), and m is the mass of adsorbents (g).
In order to determine the residual concentration of FLX in water samples during batch adsorption tests, a high-performance liquid chromatography

Kinetic Tests
Kinetic curves were obtained by collecting medium samples (500 μL) at predetermined periods (0, 5, 10, 15, 20, 25, 30, 40, 50, 60, 75, 90, 120, and 150 min). Tests were carried out at RT with an initial FLX concentration of 50 ppm. Concentrations of FLX in aqueous phase were determined using an HPLC-UV DAD system. All tests were performed in triplicate to provide mean and standard deviation values. Adsorption kinetic models are used to characterize the adsorption process (e.g. the movement of the adsorbate to the adsorbent's external surface, the internal diffusion of the adsorbate to the active sites, and the real sorption to the adsorbent's external surface) [47] [48]. Two models are mostly used in kinetics studies by researchers. Hence, it was decided to use nonlinear forms of pseudo-first-order (PFO) (Equation (4)) and pseudo-second-order (PSO) (Equation (5)) models in order to get the best-fit experimental data.
where Q e is the amount adsorbed (mg/g) at equilibrium, Q t is the amount adsorbed (mg/g) at time t (min), k 1 is the PFO adsorption rate constant (min −1 ), and k 2 is the PSO adsorption rate constant (g/g min).
Kinetic models are typically used to examine the mechanism involved during the adsorption process, as well as the location at the surface of adsorption, the chemical reaction involved, and/or the processes of diffusion. As the process of prevalence, the PFO model assumes a physical adsorption, while a chemical adsorption is inferred by the PSO model [49]. Kinetic parameters (maximum adsorption potential and kinetic constant) were determined for both models using software: the kinetic equation solve feature for Microsoft Excel, and nonlinear regression analysis (MATLAB).

Characterization Techniques of CS and N,O-CMCS
N,O-CMCS was synthesized by alkalization followed by carboxylation as shown in Scheme 1. First, CS was treated with an alkaline solution of isopropanol using NaOH, which acted as a swelling agent. The latter enables the penetration of the relatively unreactive CS polymer. Then, the CS was treated with monochloroacetic acid to generate N,O-CMCS prior to its neutralization with HCl (37%) [28] [35]. Infrared (IR) and 1 H-NMR spectroscopy provided evidences of the successful carboxymethylation and the presence on resulting spectra of distinctive CS and CMCS functional groups [35]. As reported in the literature, this reaction does not go to completion and certain hydroxyl and amino moieties remain unsubstituted [39]. Furthermore, it must be taken into account that if CS is not fully deacetylated, certain units of glucosamine and acetylglucosamine from the partial deacetylation of the parent chitin may also interfere with the reaction [50] [51]. According to Du and Hsieh (2008), the DS and yield of CMCS should be higher with longer reaction times of both alkalization (2 -12 h) and carboxylation (2 -5 h) [31]. However, they only improved at higher temperature of alkalization [31]. Based on these findings, we determined the best chemical reaction conditions and optimized the chemical synthesis of N,O-CMCS with DS value of 1.15 (≥85%). The carboxylation and alkalization were both conducted at 60˚C. During the carboxylation reaction (4 h), no additional gain was observed in terms of DS at extended time. However, increasing the reaction time of the alkalization from 2 h to 12 h had a tremendous impact on the DS. As expected, the resulting N,O-CMCS biomaterial is soluble in distilled water at 60˚C [52] [53] [54].

Zeta Potential Results
The Zeta potential (ZP) is used to measure the effective electric charge on the membrane's surface; it can provide useful information on possible electrostatic interactions between N,O-CMCS and FLX molecules. Based on the degree of Journal of Materials Science and Chemical Engineering

Electrospinning Parameters and Stabilization of Nanofiber Mats
Two of the most critical parameters affecting the electrospinning process are the concentration and the solution's mass ratio N,O-CMCS/PEO, since they play a key role in having defect-free nanofibers. Electrospun N,O-CMCS/PEO nanofiber mats were prepared from aqueous solutions at various N,O-CMCS concentrations and CMCS-PEO mass ratios. PEO was used as a co-polymer agent to facilitate the electrospinning process of CMCS. Electrospun N,O-CMCS/PEO mats were more stable with smaller nanofiber diameters than those obtained with CMCS dissolved in 1% acetic acid solution. They were also beads-free. In the present study, an optimization was conducted to find out as well the best individual polymers concentrations and right mass ratios of N,O-CMCS/PEO that  (Table S1). Electrospinning of aqueous N,O-CMCS/PEO solutions at low concentrations (CMCS/PEO 2.5 wt%:1.5 wt%) brought droplets and unstable jet during certain experiments. By contrast, when both CMCS and PEO concentrations were prepared at higher concentrations (CMCS/PEO 8 wt%:8 wt%), nanofibers with beads and large diameters were obtained. Therefore, our findings proved that the aqueous mixture composition had an important effect on the morphology and diameter distributions of electrospun nanofibers.
A comparative study of cumulated results from manufactured nanofibers of unmodified CS and modified CS (N,O-CMCS) was also completed. Figure 5 shows SEM images of one CS/PEO electrospun membrane. The material was synthetized using mixed solutions made of CS 2.5% and PEO 1.5% (4:1). The CS polymer was dissolved in 90% acetic acid, and the experimental parameters were as follows: flow rate 0.2 mL/h, distance tip-collector 7 cm, and voltage 10 kV.
Electrospinning was carried out at RT and relative humidity ranging between 16% -20%. Thus, nanofibers of pure CS (140 ± 53 nm) were produced successfully with slight modifications of our previous developed protocol [43].
The optimal aqueous mixture (N,O-CMCS/PEO) was found to be the 4:3 ratio (Table S1). Under such experimental conditions, the best fibers formation was obtained (176 ± 40 nm). It is important to mention that the formation of nanofibers is greatly affected by humidity. Indeed, nanofibers were very thin and developed inconsistently at lower humidity levels. However, an electrospraying effect was found when the humidity was higher than 35%. During our optimiza-  Optimal heating stabilization conditions of nanofibers were found to be 140˚C for 30 min. Under these conditions, membranes were stiffened, and fibers were strengthened without any surface modification. In addition, it was observed that the thermal stability was even higher, and the N,O-CMCS membrane was more water-resistant [31]. Many attempts were made to get stable nanofibers in water via different chemical treatments ( Figure 6). In all cases, the tested chemical additive was not suitable; they destroyed or altered membranes rather than stabilize them. Therefore, thermal-cross-linking stabilization of membranes at 140˚C for 30 min was found the safer and optimal way to stabilize the N,O-CMCS/PEO nanofibers without any surface damage or modification (Figure 7).

Adsorption Test and Kinetic Studies
As shown in Figure 8, To investigate the potential rate-controlling steps involved in the adsorption of FLX onto N,O-CMCS/PEO nanofibers, PFO and PSO kinetic models were used to fit the experimental data. Figure 9 shows the curves for both fitting models. Kinetic parameters are summarized in Table 2 for both models obtained by nonlinear curve fitting of experimental data with MATLAB software to     reduce statistical discrimination bias [56]. The correlation coefficients (R 2 ) of the pseudo-first-order and pseudo-second-order curves at pH 4.4 were respectively 0.9987, 0.9863. However, at pH 8.0, calculated values of R 2 were 0.9925 and 0.9949. The higher rate of adsorption (Q t = 79.7 ± 7.9 mg/g) of FLX on N,O-CMCS nanofibers was obtained at pH 8.0. Based on R 2 , kinetics curves at pH 8.0 followed a PSO rather than a PFO model. Therefore, a PSO kinetics might imply that the interaction between adsorbate and nanofibers is proceeding via chemisorption (e.g. electrostatic and ionic bonds). This type of interaction is in agreement with the experimental conditions carried out under a pH of 8.0. Precisely, N,O-CMCS nanofibers are presenting a negative ionized form (pKa COOH = 4.5), while FLX recognized as a weak base is protonated with a positive charge (pKa: 9.8). Interestingly, when the pH of the solution was 4.4, a PFO kinetic model was obtained revealing a process that could be much more governed by physical interactions (e.g. van der Waals and hydrogen bonds) at the surface of the modified biopolymer. The non-ionized form of the modified biopolymer is dominant along with its hydrophobic character (Figure 4). Accordingly, the removal yield of FLX under acidic condition is falling to 54%.
During batch adsorption tests, pH values ranged from 2 to 10. The maximum Journal of Materials Science and Chemical Engineering adsorption capacities of the adsorbent in acidic solution (pH = 4.4) were around 53.60 mg/g, in comparison to 79.7 mg/g when the pH was adjusted to a value of 8.0. The pH change of the medium had a direct impact on the rate of adsorption.
Despite the good results obtained in this study, additional tests (e.g. temperature changes, weight of adsorbent, determination of isotherms and thermodynamics) are required: 1) to better understand the sorption mechanisms at the surface of the N,O-CMCS nanofibers, and 2) to improve its efficiency in terms of absorption rate. Finally, other tests will be carried out on the reusability of the new adsorbent with consecutive adsorption/desorption cycles.
Serious challenges to meet increasing demands for clean water resources have been driving advances in technology including the use of low cost, abundant, and "green" adsorbent biomaterials. In the last decade, some articles were published on the removal of FLX in water by adsorption using different adsorbents manufactured from materials. A list of adsorbents along with their respective adsorption capacity is provided in Table 3

Conclusion
The study presented an optimized methodology for the synthesis of N,O-CMCS/PEO electrospun nanofibers (176 ± 40 nm) used as adsorbent. The developed membranes exhibited an excellent ability for the removal of FLX from water at pH 8.0 (adsorption capacity up to 79.7 ± 7.9 mg/g). Kinetics tests performed on N,O-CMCS/PEO nanofibers under optimized conditions gave a better correlation with the PSO model. Data analysis is suggesting a possible chemisorption mechanism between FLX and the N,O-CMCS/PEO nanofibers. However, further tests (e.g. isotherms, thermodynamic) will be attempted in the near future to better understand the adsorption mechanism at the surface of the modified biomaterial. From promising results obtained so far during our experiments, and a comparison made with other existing sorbents, N,O-CMCS/PEO nanofibers are believed to be efficient and suitable to remove pharmaceutical residues such FLX in water.