**Current Trends and Perspectives in the Application of Polymeric Materials for Wastewater Treatment**

Editors

**Marta Otero Ricardo N. Coimbra**

MDPI • Basel • Beijing • Wuhan • Barcelona • Belgrade • Manchester • Tokyo • Cluj • Tianjin

*Editors* Marta Otero Environment and Planning CESAM University of Aveiro Aveiro Portugal

Ricardo N. Coimbra Environment and Planning University of Aveiro Aveiro Portugal

*Editorial Office* MDPI St. Alban-Anlage 66 4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal *Polymers* (ISSN 2073-4360) (available at: www.mdpi.com/journal/polymers/special issues/curr tren pers appl poly mate wastr treat).

For citation purposes, cite each article independently as indicated on the article page online and as indicated below:

LastName, A.A.; LastName, B.B.; LastName, C.C. Article Title. *Journal Name* **Year**, *Volume Number*, Page Range.

**ISBN 978-3-0365-2372-9 (Hbk) ISBN 978-3-0365-2371-2 (PDF)**

© 2021 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license, which allows users to download, copy and build upon published articles, as long as the author and publisher are properly credited, which ensures maximum dissemination and a wider impact of our publications.

The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons license CC BY-NC-ND.

### **Contents**



### **About the Editors**

#### **Marta Otero**

Marta Otero received her Ph.D. in Chemical Engineering from University of Leon (Spain) in ´ 2000, then occupying several lecturer and researcher positions, including a Marie Curie fellowship at the Faculty of Engineering of the University of Porto (FEUP, Portugal) and a Ramon y Cajal contract ´ at University of Leon (Spain). In 2017, she moved to University of Aveiro (Portugal), where she ´ is currently senior researcher at the Associate Laboratory CESAM (Centre for Environmental and Marine Studies) and Department of Environment and Planning. Her research is mainly focussed on two fields: (i) sustainable treatments and materials for water decontamination; and (ii) bio-wastes management and valorization.

#### **Ricardo N. Coimbra**

Ricardo N. de Coimbra is a chemical engineer graduated at the Faculty of Engineering of the University of Porto (FEUP, Portugal) in 1998. Since then, his professional activity has always been related to polymers, polymeric materials and their processing. In these fields, he has carried out engineering, consulting, research and lecturing activities. In addition, wastewater treatment was the focus of his doctorate studies, which allowed him to obtain his Ph.D. at the University of Leon (Spain) ´ in 2017. After that, he has worked both in academia and industry.

### **Preface to "Current Trends and Perspectives in the Application of Polymeric Materials for Wastewater Treatment"**

Water is indispensable to the functioning of most known life forms, and good water quality is essential to human health, social and economic development, and ecosystem functioning. Nonetheless, population growth has been leading to the degradation and depletion of fresh water resources around the world. Under these circumstances, ensuring sufficient and safe water supplies for everyone is one of the Sustainable Development Goals (SDGs) set by the United Nations General Assembly in 2015 for the year 2030. For this goal to be achieved, the development and implementation of appropriate and efficient wastewater treatments that allow us to reduce water pollution is a major challenge.

The application of polymers and polymeric materials in wastewater treatment is a research field that has largely developed. Conventional and novel approaches have been carried out by researchers from different areas, who have demonstrated that polymers and polymeric materials may have an important role in the removal of pollutants of different origin and nature from wastewater, in the disposal of sludge, in the recycling of materials, in the improved efficiency and economy of wastewater, etc.

In view of the relevant contribution that polymers and polymeric materials may have in the conservation of the aquatic environment, namely by their application in wastewater treatment, original research and review papers on "Current trends and perspectives in the application of polymeric materials for wastewater treatment" were here brought together. Authors of the here included works are deeply acknowledged for their outstanding contributions, which we hope may be helpful and inspiring for readers interested in this topic.

> **Marta Otero, Ricardo N. Coimbra** *Editors*

### *Editorial* **Current Trends and Perspectives in the Application of Polymeric Materials to Wastewater Treatment**

**Ricardo N. Coimbra <sup>1</sup> and Marta Otero 1,2,\***


Water with the necessary quality is indispensable to the functioning of most of the known life forms, being essential to human health, social and economic development, and ecosystems functioning. However, only 2.5% of all water on Earth is freshwater, and less than 1% is accessible, its availability being actually affected by climate change and direct human impacts. Furthermore, population growth and industry expansion have been leading to the continuous degradation of freshwater quality around the world. Under these circumstances, ensuring sufficient and safe water supplies for everyone is one of the Sustainable Development Goals (SDGs) set by the United Nations General Assembly in 2015 for the year 2030. For this goal to be achieved, the development and implementation of appropriate and efficient wastewater treatments is a major challenge.

The application of polymers and polymeric materials in wastewater treatment is a research field that has greatly developed from the end of the last century. The very nature, structure, and versatility of polymers make them useful for many applications, including wastewater treatment processes. Conventional and novel approaches have been elaborated or refined by researchers from different areas, who have demonstrated that polymers and polymeric materials may have an important role not only in the removal of pollutants of different origin and nature from wastewater but also in the recycling of materials and the improvement of wastewater efficiency and economy.

In view of the relevant contribution of polymers and polymeric materials to the conservation of the aquatic environment, namely by their application in wastewater treatment, this Special Issue (SI) was launched for the publication of original research or review papers within this topic. The aim was to bring forth the challenges and discuss current trends and perspectives in the utilization of polymers and polymeric materials—either synthetic or natural—for the treatment or purification of wastewater.

Eleven research works [1–11] by distinguished international authors were published within this SI on "Current Trends and Perspectives in the Application of Polymeric Materials for Wastewater Treatment", which covered a wide range of issues related to this topic.

Abdulsalam et al. [1] developed hydrophilic hybrid polyvinylidene fluoride (PVDF) polyethylene glycol (PEG) ultrafiltration membranes loaded with Nano-MgO (NMO) by using a phase inversion technique. These authors demonstrated that NMO incorporation not only improved the membranes' antifouling properties and permeation performance but also allowed for an enhanced color separation from palm oil mill effluent, which was related to the synergism between surface deprotonation and pore size screening. For their part, Linhares et al. [6] loaded silver nanoparticles (AgNPs) in microfiltration polymeric membranes (15 wt.% polyethersulfone and 7.5 wt.% polyvinylpyrrolidone in *N*,*N*-dimethylacetamide) by the sputtering technique. They demonstrated the efficiency of this technique to load AgNPs, which provided membranes with biocidal properties resistant to biofouling and made them proficient for water disinfection treatments.

**Citation:** Coimbra, R.N.; Otero, M. Current Trends and Perspectives in the Application of Polymeric Materials to Wastewater Treatment. *Polymers* **2021**, *13*, 1089. https:// doi.org/10.3390/polym13071089

Received: 18 March 2021 Accepted: 24 March 2021 Published: 30 March 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/).

Mia et al. [8] loaded iron on waste silk fibers previously grafted with polydopamine by oxidative polymerization to produce wSF-DA/Fe, which was tested for the catalytic removal of toxic dyes (Methylene Blue, Cationic Violet X-5BLN, and Reactive Orange GRN). The authors postulated that the dye removal was due to the synergistic effect of free radicals and reactive species, which resulted from a heterogeneous Fenton reaction and oxidized the dyes into colorless nontoxic substances. The catalytic performance of wSF-DA/Fe was affected by the H2O<sup>2</sup> concentration, initial dye concentration, temperature, and presence of electrolytes (NaCl, Na2SO4). Aiming at the catalytic removal of Rhodamine B (RB) dye, Ansari et al. [2] synthesized Fe3O<sup>4</sup> nanoparticles (NPs) and sodium dodecyl sulfate (SDS) coated Fe3O<sup>4</sup> NPs (SDS@Fe3O4) by the co-precipitation method. RB degradation was tested in presence of H2O2, H2O<sup>2</sup> and Fe3O<sup>4</sup> NPs, and H2O<sup>2</sup> and SDS@Fe3O<sup>4</sup> NPs, the latter providing an increased catalytic removal, as the SDS coating avoided the aggregation of Fe3O<sup>4</sup> and the associated efficiency reduction.

In the field of adsorptive treatments, Khan et al. [5] investigated RB adsorption by a novel solvent impregnated resin (SIR). SIR, which was produced by modifying the cationic polymeric resin Dowex 5WX8 with the solvent t-butyl phosphate, adsorbed RB mainly by electrostatic interactions and π–π bonding. The authors optimized the operational conditions, viz. pH, SIR dosage, and contact time, for the adsorptive removal of RB. The adsorptive removal of Cr(VI) from water was studied by Yang et al. [11], who produced for this purpose activated carbon microspheres (SLACM), using sodium lignosulfonate ((SL), a waste from the pulp and paper industry) as raw material. The synthesis of SLACM, which was to be attained with no binder addition, consisted of (i) amination of chloromethylstyrene-divinylbenzene-styrene copolymer (CMPS) with 1,3-diaminopropane; (ii) Mannich reaction of SL and amino CMPS to produce the adsorbent resin microsphere; (iii) impregnation with ZnCl<sup>2</sup> and pyrolysis at 600 ◦C. The adsorption of Cr (VI) by the produced SLACM was shown to be favored by decreasing pH (within pH 2 and 9) and increasing temperature (within 20 and 40 ◦C). Hoshima et al. [4] aimed at the adsorption of the rare earths dysprosium (Dy) and neodymium (Nd) from water as a previous necessary step for their subsequent recovery. With this objective, the authors produced a new adsorbent by the radiation-induced graft polymerization of methacrylate with a long alkyl chain on a PE/PP nonwoven fabric and the subsequent loading of 2-ethylhexyl hydrogen-2-ethylhexylphosphonate by hydrophobic interaction and chain entanglement between the alkyl chains. Four different methacrylate monomers were tested, namely butyl methacrylate (BMA), hexyl methacrylate (HMA), dodecyl methacrylate (DMA), and octadecyl methacrylate (OMA). Only the OMA-adsorbent was stable under subsequent uses, which was related to the suppression of EHEP losses due to the strong hydrophobic interaction and chain entanglement between the long alkyl chains. Moreover, OMA-adsorbent was efficient in the adsorption of Dy (III) and Nd (III) from water, being a promissory material for the recovery of rare-earth metals from NdFeB permanent magnet scraps.

Four papers in the SI dealt with the adsorptive removal of pharmaceuticals [3,7,9,10]. Mashile et al. [7] synthesized a magnetic mesoporous carbon/-cyclodextrin–chitosan (MMPC/Cyc-Chit) nanocomposite for the adsorption of fluoroquinolones (FQs), viz. danofloxacin, enrofloxacin, and levofloxacin, from different water samples, including a synthetic mixture of the FQs, the influent and effluent from a wastewater treatment plant, river water, and tap water. The authors found that incorporating biodegradable polymers such as chitosan and cyclodextrin into magnetic mesoporous carbon brought about a nanocomposite with large surface area and adsorption capacity. This was a very complete study, which included the optimization of operational parameters (pH, mass of MMPC/Cyc-Chit, and sonication power) by a response surface methodology, kinetic and equilibrium modeling, assessment of regeneration and reusability, thermodynamic parameters determination, and cost analysis for the synthesis of MMPC/Cyc-Chit. Mohammadi et al. [9] synthesized a poly(styrene-block-acrylic acid) diblock copolymer/Fe3O<sup>4</sup> magnetic nanocomposite (P(St-b-AAc)/Fe3O4)) and tested it for the adsorption of antibiotic ciprofloxacin from synthetic wastewater. This work included the optimization of

operational parameters, namely antibiotic concentration, pH, nanocomposite mass, and contact time, and showed that P(St-b-AAc)/Fe3O<sup>4</sup> was efficient in the adsorptive removal of ciprofloxacin. Aiming at the adsorption of selective serotonin reuptake inhibitor (SSRI) antidepressants and their metabolites from water, Gornik et al. [3] optimized the synthesis of molecularly imprinted polymer (MIP) adsorbents in which the SSRI sertraline was used as template. Different MIPs were synthesized by varying the functional monomer, the porogen, and/or the template form, which were shown to largely affect the adsorbent performance of the resulting material, so the authors selected the most efficient. Even when the selected MIPs had a relatively lower surface area than conventional activated carbons, they displayed a larger adsorption capacity in real wastewater samples. Apart from the MIPs optimization and characterization, this work included the assessment of their reusability, occurrence of cross-reactivity, adsorption kinetics, matrix effects, upscale, and leaching, with authors pointing out the necessity of carrying out future work at a larger scale to confirm the advantages of the synthetized materials. In order to overcome the drawbacks of MIPs, mainly the small number of recognition sites per unit of volume and the low mass transfer, surface molecular imprinting on magnetic yeast (MY) was carried out by Qiu et al. [10], who synthesized highly selective magnetic yeast-molecularly imprinted polymers (MY@MIPs) for the adsorptive removal of antibiotic sulfamethoxazole (SMX). For the production of MY@MIPs, these authors started by preparing nano-Fe3O<sup>4</sup> by an in situ one-step procedure, which was loaded onto yeast cells to obtain MY. Then, MY was used as core for the polymerization of MIPs using SMX as template to produce MY@MIPs. The authors compared MY@MIPs with MY and non-imprinted MY@NIPs (synthesized in the same way as MY@MIPs but in the absence of template) and evidenced the superior SMX adsorption capacity of MY@MIPs. Furthermore, besides the selectivity of MY@MIPs towards SMX in the presence of other pharmaceuticals and in real wastewater, the reutilization capability of this material was also proved, pointing to its possible application as an alternative adsorbent for SMX selective removal from wastewater.

Membrane filtration and catalytic applications together with the utilization of polymeric materials for the adsorptive removal of pollutants were discussed in this SI. Furthermore, it is worth highlighting the attention given to the adsorptive removal of pharmaceuticals from wastewater. Since the 1990s, a growing scientific concern about pharmaceuticals may be inferred from the number of related publications, and this SI was no exception. Such a concern is mainly related to (i) the analytic development that has made possible their detection at trace levels and the confirmation of their ubiquity in environment; and (ii) the ecological risks related to their potential to cause physiological responses in nontarget individuals, including endocrine disruption and antibiotic resistance. An important benefit of pharmaceuticals removal by adsorption is that such a treatment does not result into the formation of by-products, which in some cases can be more hazardous than the parent compounds.

Polymers have been used for long in conventional wastewater treatment for the flocculation/coagulation of solids, so that they may be easily separated from water. However, the SI hereby presented makes evident that polymers and polymeric materials may have many new and varied applications in wastewater treatment. Most published works take advantage of polymers' versatility and capacity to be combined or modified to produce advanced materials with relevant features for water treatment. In fact, polymer modification is presently a hot topic, since it allows for the development of specific and even smart materials for target applications in different sectors, including wastewater treatment. Owing to the efforts of polymer engineers and scientists, many alternative and advanced polymeric materials are continuously developed. In the specific case of wastewater treatment, polymer applications have become very important due to the increased pollutant removal efficiencies that they offer, so this is a research field in great expansion. The progress in the last years and the successful applications of polymer and polymeric materials are reflected by the high quality works published within this SI.

**Author Contributions:** M.O. and R.N.C., as Guest Editors of the Special Issue entitled "Current Trends and Perspectives in the Application of Polymeric Materials for Wastewater Treatment", contributed to the preparation of this Editorial. Conceptualization, M.O. and R.N.C.; writing original draft preparation, R.N.C. and M.O.; writing—review and editing, M.O.; Supervision, M.O. Both authors have read and agreed to the published version of the manuscript.

**Funding:** Thanks are due to the Portuguese Fundação para a Ciência e a Tecnologia/ Ministério da Ciência, Tecnologia e Ensino Superior (FCT/MCTES) for the financial support to the Associated Laboratory CESAM (UIDP/50017/2020+UIDB/50017/2020) through national funds. FCT is also acknowledged for funding through the Investigator Program (IF/00314/2015).

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

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data sharing is not applicable to this article.

**Acknowledgments:** Contributions to this Polymers Special Issue (SI) are deeply acknowledged. As Guest Editors, we would like to sincerely thank the authors of these contributions for considering this SI for the publication of their outstanding works. Thanks are also due to all the anonymous reviewers who gently reviewed these works and sent their wise comments, corrections, and suggestions. Finally, our recognition to the editorial managers for their helpful assistance in the assembling of this SI.

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

#### **References**


*Article*

### **Permeability and Antifouling Augmentation of a Hybrid PVDF-PEG Membrane Using Nano-Magnesium Oxide as a Powerful Mediator for POME Decolorization**

#### **Mohammed Abdulsalam 1,2 , Hasfalina Che Man 1,\* , Pei Sean Goh <sup>3</sup> , Khairul Faezah Yunos <sup>4</sup> , Zurina Zainal Abidin <sup>5</sup> , Aida Isma M.I. <sup>6</sup> and Ahmad Fauzi Ismail <sup>3</sup>**


Received: 19 December 2019; Accepted: 7 January 2020; Published: 3 March 2020

**Abstract:** This study focused on developing a hydrophilic hybrid polyvinylidene fluoride (PVDF)-polyethylene glycol (PEG) hollow membrane by incorporating Nano-magnesium oxide (NMO) as a potent antifouling mediator. The Nano-hybrid hollow fibers with varied loading of NMO (0 g; 0.25 g; 0.50 g; 0.75 g and 1.25 g) were spun through phase inversion technique. The resultants Nano-hybrid fibers were characterized and compared based on SEM, EDX, contact angle, surface zeta-potential, permeability flux, fouling resistance and color rejection from palm oil mill effluent (POME). Noticeably, the permeability flux, fouling resistance and color rejection improved with the increase in NMO loading. PVDF-PEG with 0.50 g-NMO loading displayed an outstanding performance with 198.35 L/m<sup>2</sup> ·h, 61.33 L/m<sup>2</sup> ·h and 74.65% of water flux, POME flux and color rejection from POME, respectively. More so, a remarkable fouling resistance were obtained such that the flux recovery, reversible fouling percentage and irreversible fouling percentage remains relatively steady at 90.98%, 61.39% and 7.68%, respectively, even after 3 cycles of continuous filtrations for a total period of 9 h. However, at excess loading of 0.75 and 1.25 g-NMO, deterioration in the flux and fouling resistance was observed. This was due to the agglomeration of nanoparticles within the matrix structure at the excessive loading.

**Keywords:** nano-MgO; structural modification; permeability; antifouling; color rejection; POME

#### **1. Introduction**

Owing to the exceptional physical, mechanical and chemical stability of polyvinylidene difluoride (PVDF) polymer, its applications for bio-system/tissue engineering [1], pervaporation [2,3] and separation technology [4,5] have gained a considerable attentions. In term of solubility, PVDF polymer can easily dissolve in most of the organic solvents such as *N*,*N*-dimethylformamide (DMF), *N*,*N*-dimethylacetatamide (DMAc), Triethylphosphate (TEP) and *N*-methyl-*N*-pyrolidinone (NMP) [6,7]. As a result of this flexibility, PVDF polymer is suitable for fabricating polymeric membrane [6]. It has likewise been noted that PVDF membrane requires a relatively low pressure and minimal energy demand during filtration. This feature is most desirable for the microfiltration and ultrafiltration separation processes [8]. The aforementioned separation process using PVDF polymeric membrane have been applied on several wastewater, such as palm oil mill effluent [9], dye wastewater generated from textile industry [10], saline-water [11] and endocrine compounds [12]. Some drawbacks were reported which include continuous diminishing in permeation, low rejection as well as fouling [13]. The reduction in the membrane flux and susceptibility to fouling was collectively attributed to the hydrophobic nature of the polymer [14,15]. Thus, this justified the significant research conducted to subdue the hydrophobicity by improving the water-liking property along with permeability flux of the polymeric membrane using inorganic nanoparticles as an additive [16].

The commonly applied nanoparticles to modified polymeric membranes includes ZnO [17,18], TiO<sup>2</sup> [9], Ag2O<sup>3</sup> [19,20], Al2O<sup>3</sup> [21], graphene oxide [22,23], SiO<sup>2</sup> and CuO [24]. Tan et al. [17] enhanced the permeation and antifouling properties of a bared PVDF membrane by incorporating synthesized Zn-Fe oxide (ZIO) into the matrix structure. The nanocomposite membrane with 0.5 wt % ZIO loading demonstrated a significant increase in permeate flux to the magnitude of 25% increment compare to the bared PVDF membrane. Also, Shen et al. [25] modified the matrix structure of polymeric membrane using ZnO nano-additive. The result showed that the additive has sizeable influences on the pores structure as well as the hydrophilicity of the membrane. Over 254% improvement in permeability flux was reported at the loading of 0.3 g Nano-ZnO. Furthermore, the authors validated that the porosity of the modified composite membrane increased with the presence of the nano-additive, which ultimately justify the reported high flux. Nano-ZnO has large surface area that expedite the formation of hydroxyl (–OH) functional group on the surface [26,27]. The presence of –OH enhanced hydrophilicity of the PVDF polymeric membrane, thus mitigating fouling rate [26]. Analogously, TiO<sup>2</sup> Ag2O<sup>3</sup> and Al2O<sup>3</sup> also exhibit similar features when incorporated into the matrix structure of polymeric membrane. Subramanium et al. [9] synthesized titanate nanotubes using nano-TiO<sup>2</sup> as a precursor to modify the PVDF membrane. The results substantiated that under photo-catalytic condition, the involvement of TiO<sup>2</sup> in the dope formulation improved the permeation consistency (35.8 L/m<sup>2</sup> ·h) of the resultant nanocomposite membrane. Negligible fouling was observed at 0.5 wt % TNT loading throughout the 4 h continuous filtration. Another report has shown strong agreement with this observation, and the results affirmed that the antifouling performance of TiO<sup>2</sup> was due to generated –OH under UV spectrum and its aptitude to exhibit self-cleaning [28]. Also, attentions are long drifted to the use of nano-Ag2O<sup>3</sup> to modify polymeric membrane [19,20,29]. The nano-additive released Ag ions to inhibit the metabolism of the microbial-foulants, thereby preventing the generation of the extracellular polymeric substance (EPS). Thus, this ultimately curtailed the most serious type of fouling, which are organic and bio-fouling [22]. Maximous et al. [30] investigated the antifouling impact of Al2O<sup>3</sup> and the results showed that at optimum 0.05 wt % loading, the modified composite polymeric membrane was less prone to fouling. Besides, the presence of Al2O<sup>3</sup> in the matrix structure of a polymeric membrane not only reduces the fouling but also improves the flux consistency.

However, most of the widely used nanoparticles, particularly as mentioned above, are photo-catalytic driven to effectively address the fouling issue [31,32]. This implies that the presence of ultraviolent radiation is a prerequisite to precede the antifouling performance. More so, the issue continues releasing antifouling radicals and superoxide could seriously jeopardize the stability of the composite matrix structure. In line with this, Tan et al. [17] reported that the structural instability of the modified nanocomposite PVDF-ZIO membrane was discernible after four filtration cycles due to the collapse of the incorporated nanoparticles. This could significantly undermine the overall antifouling performance and also re-exposing the modified membrane to inconsistencies in permeation flux along with frequent fouling challenges. Moreover, the nano- ZnO, TiO2, Ag2O3, Al2O<sup>3</sup> and CuO are relatively expensive and the running cost could be unsustainable for industrial application [33,34]. Meanwhile, nano-MgO has remained one of the antimicrobial and super hydrophilic nanomaterials yet to be fully explored in improving filtration and antifouling performance of polymeric membrane [35]. The nano-MgO has the ability to release reactive oxygen species (ROS) which directly extract lipid from the cells of the microbial-foulants. This ultimately disrupted the metabolic activities and hindered biofilm formation [31,36]. More interestingly, the nano-MgO not only averts bio-fouling formation but also capable of evincing self-cleaning mechanism [31,36]. The nano-MgO precursor is readily available and comparably cheaper than other nanomaterials (such as TiO2, Ag2O3, Al2O3, and ZnO) [31]. Therefore, MgO-nanoparticles intimate a promising additive capable of enhancing antifouling properties and permeation performance of polymeric membranes for industrial filtration purposes. Currently, application of a modified Nano-MgO (NMO) composite PVDF-PEG membrane for separation of color pigment from POME has not been reported.

In view of these, the present study focuses on modifying the structure of an in-house fabricated PVDF-PEG ultrafiltration membrane at various loadings of NMO. The impacts of the incorporated NMO were examined based on the morphological changes, hydrophilicity, the permeability flux and fouling resistance of the resultant membrane. Furthermore, the rejection and color separation efficiency was studied, and also the used membranes were characterized using Fourier transform infrared spectroscopy. The outcome of the study demonstrated that the involvement of the NMO in the dope formulation significantly improved the antifouling properties and the permeation performance, alongside with color separation from palm oil mill effluent (POME) via the synergy of surface deprotonation and pore size screening mechanism.

#### **2. Experimental Methods**

#### *2.1. Chemicals and Materials*

Nano-MgO (NMO: particles size (BET) < 50 nm; MW = 40.30 g/Mol; purification ≥ 99.9%), Pellets PVDF (Kynar 740) and polyethylene glycol (PEG: 12,000 g/mol) were procured from Sigma Aldrich (M) Sdn Bhd, Selangor, Malaysia. The PVDF was applied as the major membrane matric polymer, while the PEG as co-polymer to enhance pore formation. The *N*,*N*-dimethylformamide (DMF: ≥99.9%; 87.12 g/mol), ethanol (≥99.98%; 46.07 g/Mol) and glycerol (≥99%; 92.09 g/Mol) were also obtained from Sigma Aldrich (M) Sdn Bhd, Selangor, Malaysia, and respectively used as doping and post-treatment solvents without any further purification. LiCl2.H2O (MW 42.39 g/Mol; ≥99%) was purchased from Acros Organic Industry, Chemicals and Reagents, Semenyih, Selangor, Malaysia and it was applied to improve the hydrophilicity of the polymers (PVDF/PEG). Also, high strength POME with initial color concentration of 8570 ADMI was collected from an Oil Palm Milling industry, Malaysia. Prior to the usage, the POME sample was filtered to remove all the visible debris and then diluted using a factor of 2 to give 4285 ADMI. This procedure was to represent the industrial final discharged color concentration ranges [9].

### *2.2. Synthesis of Nano-Hybrid PVDF*/*PEG-NMO*

#### 2.2.1. Dope formulation

The dope formulation was preceded by adding NMO into DMF and then subjected to sonication using digital ultrasonic water bath (VWR 142-0300) at 75 ◦C for a period of 20 min. Subsequently, LiCl2·H2O was added into the mixture under steady agitation of 350 rpm and 75 ◦C for a period of 24 h. Essentially, this procedure assists in dispersion of the NMO as well as ensuring good blending of the mixtures. The mixture was followed by adding the dehydrated PVDF pellets and the co-polymer, PEG. The combination was stirred continuously at 350 rpm and 100 ◦C for another 24 h using a hot-plate stirrer (Monotaro; C-MAG HS7, Malaysia) to achieve a homogenous solution. The amount of the NMO contained in the dope solutions were varied as contained in Table 1.


**Table 1.** Chemical formulation of Neat PVDF-PEG and modified Nano-hybrid membranes.

#### 2.2.2. Spinning of Nano-Hybrid PVDF/PEG-NMO Hollow Membrane

It is important to note that same spinning parameters were applied for all of the samples. The dopes were spun through dry-jet wet swirling technique employing an annular spinneret. The inner and outside diameter of annular spinneret was 0.55 and 1.15 mm, respectively. During the fabrication of the membrane, tap water and distilled water was used as the external and internal coagulant. In addition, the pick speed control, collecting drum speed, extrusion rate, air gap, external coagulant temperature, room temperature and room humidity remains constant at 7 rpm, 11 rpm, 5 mL/min, 10 cm, 25 ◦C, 29.5 ± 1 ◦C and 72.7%, respectively. The spun fibers were drench in a continues flow water-bath at least for a period of 24 h to dislodged solvent remnants. In order to minimize shrinkage, post-treatments were applied on the fibers by immersing in ethanol for 12 h, then followed by drenching in glycerol for 5 h, respectively. The post-treated fibers were air-dried for at least 24 h to ensure complete dehydration.

#### *2.3. Characterization of Nano-Hybrid PVDF*/*PEG-NMO Fibre*

#### 2.3.1. Morphology Analysis

The cross section and surface morphology of the synthesized Nano-hybrid fiber were examined using scanning-electron-microscope, (SEM: S-3400N). Erstwhile, the composite fibers were immersed into liquid nitrogen for a period of 5 min. This procedure ensures sharp breaking of the fiber to reveal the cross sectional structure. Then, the fractured samples were sputtered coated with gold thin layer, and the voltage acceleration was maintained constant at 20 kV during the image capturing.

Also, SEM/EDX (scanning electron microscope/energy dispersive X-ray) was engaged for examining the NMO particle distribution and the dimension, as well as elemental composition in the Nano-hybrid fiber. The fractured membrane samples were place on the adhesive carbon tape of a metal plate and examined at 20 kV accelerating voltage using SEM-Thermo Scientific (Hitachi & S-3400N) for the analysis and imaging.

#### 2.3.2. Hydrophilic and Porosity Analysis

The hydrophilicity of the synthesized membrane was analyzed based on static contact angle using goniometer (GmbH OCA 15pro, Data-Physics). The goniometer uses RO water as the probing liquid, such that the liquid droplets were captured with the equip camera after 30 s stand. The measured contact angles were analyzed using SCA20 software. In order to minimize error, each of the measurement was repeated ten times on different spots of the membrane and the average was determined as the contact angle. Throughout the analysis, the measurements were conducted at ambient temperature.

The procedure employed for the determination of porosity is based on gravimetric method [9]. Ten pieces of the synthesized membrane of 20 cm of equal length were hermetically sealed at both ends using glue-resin. The prepared samples were submersed in distilled water for 5 h under ambient temperature and humidity. Afterwards, the superficial water drops on the surface of the samples were eliminated using dry tissue paper, and then weighed as wet membrane (*M*w). The wet membrane

samples were air dried overnight at 70 ◦C and reweighed as dried membrane (*M*d). Hence, the porosity (ε) for each sample was determined using Equation (1):

$$\text{1\ }\varepsilon\text{.}\ (\%) = \frac{1}{\rho\_{\text{W}}} \times \left(\frac{M\_{\text{W}} - M\_{\text{d}}}{V}\right) 100\tag{1}$$

where ε Symbolises the membrane porosity in %, ρ<sup>w</sup> is density of water, (*M*<sup>w</sup> − *M*d) is the quantity of pores water in gram and *V* is the volume of the membrane sample.

#### 2.3.3. Surface Charge Analysis

The surface charge of the membranes was examined based on the potential of streaming analysis conducted using Anton SurPass Analyser (Paar Inc. Ireland). The samples were placed on the sample-holder by mean of a carbon-adhesive tape, and then the analyzer was pre-set to a maximum pressure of 400 mbar. This procedure was to ensure laminar flow throughout the analysis [17]. Also, a solution of 1 mM KCl was applied as contextual electrolyte and the pH varied between 2 and 10 was accomplished using aqueous 0.1 M HCl or NaOH. The Helmholts Smoluchowski procedure was adopted for the calculation of the Zeta-Potential surface charge.

#### *2.4. Membrane Performance*

#### 2.4.1. Permeability Analysis

The filtration performance was examined using a fabricated dead-end permeation system, equipped with membrane module cell and a peristaltic pump to provide the required suction pressure. Each of the modules comprises of 10 units of the membrane with equal length of 20 cm. initially, the membrane was compacted at a pressure of 0.4 MPa for a period of 30 min to ensure steady flux, while the subsequent filtration operations were performed at lower pressure of 0.3 MPa. The pure water (*J*w) and permeates flux (*J*p) were determined using Equation (2):

$$J = \frac{V}{A\_s \Delta t} \tag{2}$$

where *J* denotes the flux in L/m<sup>2</sup> ·h, *V* is the volume of permeate (L), *A*<sup>s</sup> is membrane surface area (m<sup>2</sup> ) and ∆*t* is filtration time in h.

#### 2.4.2. POME Decolorization

The membrane fibers were further subjected to filtration of diluted POME with 4285 ADMI color concentration using same set-up as applied when pure water was used as feed. Initially, the membranes fibers were engrossed in the POME solution for 90 min to initiate adhesion of thin layer color pigments on the fibers. This procedure assists in achieving accurate color rejection performances of the fibers [9]. The apparent color content of the feed POME and permeate were analyze in ADMI using UV-spectrophotometric technique (DR4000U, HACH) at an absorbance wavelength of 400 nm. The POME decolorization efficiency was determined using Equation (3):

$$\% \text{ C}\_{\text{removal}} = \left( 1 - \frac{\text{C}\_{\text{Permeate}}}{\text{C}\_{\text{Feed PONE}}} \right) \times 100 \tag{3}$$

where, *C*remova*<sup>l</sup>* is percentage of color removal in %, *C*permeate is permeate color concentration in ADMI, and *C*Feed POME is feed POME color concentration in ADMI.

#### 2.4.3. Fouling Analysis

In order to mimic practical situation and to evaluate the reusability, the membranes were subjected 3 cycle of continues filtration for a total period of 9 h using a known color concentration (4285 ADMI) of a diluted POME as feed. In each of the completed cycle, the antifouling performances of the membrane were evaluated after 180 min continues filtration without interruption. The indices used for the antifouling analysis were as stated in Equations (4)–(6). At every completion of each filtration cycle, the volume of POME permeates, flux and color rejected were determined. Then, the used membranes were physically cleaned under running tap-water for a period of 15 min. The washed and cleaned membranes were reapplied for the 2nd cycle of POME filtration for another 180 min, following same procedure as highlighted above. In this study, the POME filtration cycle was repeated 3 times to determine the antifouling performance of the membranes using percentage of flux recovery (*FR*), reversible fouling (*RF*) and irreversible fouling (*IF*) as indices Equations (4)–(6):

$$\text{Percentage of flux recovery}\_{\prime} \text{ (\%FR)} = \frac{I\_{\text{W2}}}{I\_{\text{W}}} \times 100\tag{4}$$

$$\text{Percentage of reversible following, }\left(\% \text{RF}\right) = \% \text{FR} - \frac{f\_{\text{P}}}{f\_{\text{W}}} \times 100\tag{5}$$

*Percentage o f irreversible f ouling*, (%*IF*) = (1 − %*FR*) × 100 (6)

where, *J*w2 is the water flux after the POME filtration, L/m<sup>2</sup> ·h.

#### 2.4.4. Characterization of used Membranes by FTIR Analysis

The surface chemical functional groups and transformation of the neat and modified membrane (0.5 g-NMO) after used were characterized using Fourier transform infrared spectroscopy (FTIR-Perkin Elmer spectrum 100 Series). The spectra analysis was taken over a wide range from 400 to 4000 cm−<sup>1</sup> . Essentially, the analysis involves shining a beam of light rays with variable frequencies, and then measures how much of that rays got absorbed by the specimen (i.e., the membrane samples). This ensure high signal-to-noise ratio of the spectra; thus accurate analysis of fouling level based on functional group is achievable [37].

#### **3. Results**

#### *3.1. E*ff*ect of Nano-MgO on Membrane Characteristic*

#### 3.1.1. Morphological Studies

Figure 1 presents the scanning electron microscope (SEM) images of cross-sectional view of the neat and modified hybrid Nano-MgO (NMO) PVDF-PEG membranes. It is obvious that the neat fiber had 3 distinctive layers with thin layers both at inner and outer section of the membrane. The middle layer constitutes majorly of sandwich-like morphology containing short finger-like pores at both sides toward the ultra-thin layers. However, different scenarios were observed with the modified Nanocomposite membranes. With the increasing NMO loading, the finger-like pores became longer and the number of the micro-pores structure increased considerably Figure 1b–d [38]. In addition, some spongy macrovoids were also noticed towards the inner ultra-thin layer at the higher NMO loading. This observation is in agreement with previous studies reported [38–40]. Though at 1.25 g-NMO loading, the pores structure was significantly suppressed and this may be due to the inhomogeneity dispersion of the Nanoparticles [25]. The uneven dispersion of the NMO at higher dosage resulted in the formation of agglomerated particles within the matrix structure [41]. Thus, the resultants membrane comprised of dense structure with suppressed pore sizes as shown in Figure 1e. From Figure 1b,c, homogenous dispersion of NMO can be observed, and this indicates a good compatibility at the loadings within the matrix structure.

**Figure 1.** *Cont.*

**Figure 1.** SEM Micro-structure of cross-sectional view of the Spun Fiber: (**a**–**e**) with different NMO loading.

In recap, NMO loading has demonstrated a notable effect on the shape and magnitude of the pores formation. An increase in the dosage of NMO results in continuously suppression of the sandwich structure and accompanied with the increase of spongy-finger like pores at the inner and outer walls of the fiber, respectively, as indicated in Figure 1b–d. However, a contrary scenario was observed in the Figure 1e, which presented dense structure even with higher NMO loading of 1.25 g. This is indicating that a relatively uniformly distributed NMO within the polymeric matric structure were achieved at a loading range of 0.25–0.75 g to give larger surface interaction with O–H in the coagulating water bath [42]. This phenomenon often leads to formation of finger-like pore substructure and/or combination with spongy voids, fine gravimetric, porosity and thin skin layers [43]. Conversely, the tendency of non-uniformly distribution of the augmented nanoparticles (NMO) advances with increase in the dosage (1.25 g), since the spinning parameters were maintained constant for all the samples [44]. Thus, the excessive NMO got agglomerated, thereby skewing the surface interaction with O–H during crystallization. This effect in conjunction with the increase in the viscosity of the dope solution due to the higher NMO dosage, jointly delayed the demixing process, and consequently, the formation of denser-sandwich structure alongside with suppressed finger-like pore structure [25,41,43]. In addition, high NMO content dopes are in meta-stable states that are extremely supersaturated with respect to polymer crystallization [43], and as such there may exhibit a considerable amount of pre-nucleation embryos along with several aggregated Nano-particles in the dope [25,43].

The results of the porosity analysis concurred with the morphological structure observed from the SEM characterization. Figure 2 displayed the porosity analysis of the spun neat and modified hollow fibers (a–e). The neat membrane (a) had the least porosity of 64.12%, while the modified membranes recorded higher values of 67.38%, 78.96%, 81.51% and 70.17% for b, c, d and e fiber, respectively. Similar remarks have been previously reported on the use of NMO to modify polymeric membranes [39,40,45]. The blended NMO accelerated the exchange speed between the non-solvent (water) and solvent (DMF) through the gelation process, thus, the porosity was increased considerably. The positive effect of the NMO on the porosity property could improve the membrane permeate flux significantly [45,46]. However, the diminished porosity percentage noticed with 1.25 g NMO loading might be due to the retardation in the crystallization which resulted from the high viscosity of the dope solution and presence of agglomerated particle at the excessive loading [47]. This effect resulted in the formation of a denser structure with suppressed pores and several visible aggregated particles, Figure 1e. Therefore, it can be deduced that the excessive NMO undermine the pores formation which could significantly diminish the porosity along with permeability of the resultant membrane.

**Figure 2.** Porosity properties of neat (**a**), and modified membrane with varied NMO loading (**b**–**e**).

#### 3.1.2. Membrane Hydrophilicity

The hydrophilic properties of the neat and modified membranes were examined based on surface contact angle analysis. Principally, surface with a lower contact angle is said to be more hydrophilic and water-liking [48]. Results of the contact angle analysis along with the respective goniometric images for the fabricated membranes are presented in Figure 3. Essentially, the goniometric image of the dropped probing water on the horizontally-positioned membrane samples were captured after 30 s. The membrane surface hydrophilicity increased with the increasing NMO loading. As can be seen from the Figure, the neat membrane presents the highest contact angle to the magnitude of 87.34◦ . However, considerable reductions in the contact angle were noticed with 0.50 g-NMO and 0.75 g-NMO loading to 60.01◦ and 57.19◦ , respectively. Essentially, the decrease in the contact angles is an indication of improvements in the hydrophilicity [24]. However, at higher NMO loading of 1.25 g, the contact angle upsurges to 75◦ and probably this might be due to the inhomogeneity dispersion and aggregation of the NMO particles which essentially reduced its overall effect. In a whole, the hydrophilicity of the modified membrane has shown a correlation with the dosage of the NMO, which also influences the pores morphology [40], formation of hydroxyl functional group [17,39] and surface charges [42].

**Figure 3.** Hydrophilicity of neat PVDF-PEG (**a**), and modified Nano-hybrid membranes (**b**–**e**) with the respective goniometric images after 30 second of contact.

#### 3.1.3. Elemental Analysis

EDX mapping was used to examine the presence of NMO within the matrix structure of the spun hollow fibers. Figure 4a presents the EDX spectrum of the neat membrane, and it clearly shows no trace of NMO in the composition. The major constituted elements are the C and F, which were the primary elements of the polymeric materials (PVDF/PEG). However, Figure 4b–e has evidenced the presence of the modified Nanocomposite membranes with various loadings of NMO particles. As expected, Mg composition was observed in the spectrum of all the modified fibers at distinct intensities with respect to the NMO dosage. The mapping of Mg in EDX spectra indicates successful blending of the NMO into the matrix structure as well as on the surface of the membranes. Despite the compatibility, only the 0.25 g-NMO and 0.5 g-NMO modified membranes present a free agglomeration with the matrix as depicted in Figure 4b,c. However, at higher NMO loading (0.75 and 1.25 g), an inhomogeneity dispersion and cluster of particles were observed (Figure 4d,e). Reports have shown that presence agglomerated particles in dope solution increases the viscosity, and the this results to irregular nucleation along with uneven crystallization process during the phase separation process [36].

**Figure 4.** *Cont.*

**Figure 4.** SEM/EDX spectra Mapping the Presence of NMO in (**a**–**e**) membrane.

#### 3.1.4. Surface Charge of Nano-MgO Membrane

− − − − The surface charges of the neat and modified membrane were examined using surface zeta-potential analyzer. The analysis of the surface zeta-potentials for the fabricated hollow fibers with respect to the varied wide range of pH (2 to 10) are presented in Figure 5. The equilibrium isoelectric point of the neat membrane was observed at the pH 3.6, which is in agreement with the previous study [49]. However, the addition of NMO into the dope formulation significantly influenced the surface negativity of the membranes due to the oxidation effect and the formation of the acidic oxides [17]. As shown in Figure 5, the surface of the membranes became more negatively charged with the increasing amount of NMO. This indicates that the NMO became hydrated when added into DMF solvent [17,47]. During the spinning process, the NMO exposed to the membrane surface were hydrolyzed to form some hydroxyl functional groups in the presence of water. The protonation of the NMO has resulted in the deprotonation of membrane surface [42]. Thus, the membrane surface became more negatively charged [17,25]. The fiber with 0.75 g NMO loading presented the highest negative zeta potential within the pH range of 5.5 and 9 with value of −41.99 and −57.44, respectively (Figure 5). The lower surface charges in the 1.25 g NMO loading might be due to excessive agglomeration which resulted in the uneven dispersion and reduction in overall active-surface-area [42]. The membrane with 0.5 g NMO loading recorded higher zeta potential than the fiber with 0.75 g loading, (Figure 5). From same figure, it is obvious that the neat membrane had the least zeta potential at both extreme of the pH range (5.5 and 9) with value of −18.43 and −38.17, respectively. Based on this result, it can be deduced that to derive intensive surface negativity, NMO loading should be maintained between 0.5 and 0.75 g. One of the important applications of negatively surface charged membrane is in the separation of like-charged color pigments; such as the lignin and tannin substances in a typical oily wastewater that contains POME [9]. − −

**Figure 5.** Surface zeta potential of neat PVDF-PEG, and modified Nano-hybrid membranes.

More so, researchers have reported that the main effect of the deprotonation process of a Nano-composite polymeric membrane is the resulted hydroxyl functional groups (OH–) formed on the membrane interacting surface in aqueous media, and this accounted for the surface negatively charge zeta potential [17,25]. More interestingly, the presences of this OH– on the membrane interacting interface improves its water-liking properties (hydrophilicity), permeability as well as repulsion of hydrophobic foulants, such as extracellular polymeric substance (EPS) [9,17]. This remark is strongly in agreement with other studies [17,39,46].

#### *3.2. E*ff*ect of Nano-MgO on Membrane Performance*

#### 3.2.1. Permeability

Permeability results of the spun fibers both in pure water and diluted POME (4285 ADMI concentration) are shown in Figure 6. The modified fibers demonstrated higher permeate flux performance in both pure water and the diluted-POME. The modified membrane with 0.5 g-NMO and 0.75 g-NMO loading recorded almost equal water flux with 198.35 L/m<sup>2</sup> ·h and 201.22 L/m<sup>2</sup> ·h, respectively. The 1.25 g-NMO and 0.25 g NMO modified membrane had 97.40 L/m<sup>2</sup> ·h and 83.02 L/m<sup>2</sup> ·h water flux, respectively, whereas, the neat membrane presented the least flux of 80.72 L/m<sup>2</sup> ·h. Based on the flux pattern, it shows that the addition of NMO has a positive impact on the permeability, and this may be due to its ability to enhance the hydrophilicity through the protonation process. However, this effect seems otherwise at higher NMO loading of 1.25 g, Figure 6. This may be attributed to the suppressed pores and dense structure which were the consequential effect of the excessive aggregated particles present within the matrix structure.

**Figure 6.** Flux of pure water and POME for neat PVDF-PEG (**a**), and modified Nano-hybrid fibers (**b**–**e**).

–

" " Similar trend was observed in the filtration of the diluted-POME. Both 0.5 g-NMO and 0.75 g-NMO modified membranes presents higher flux of 61.33 L/m<sup>2</sup> ·h and 57.97 L/m<sup>2</sup> ·h, respectively. On the contrary, the neat membrane had the least flux of 15.13 L/m<sup>2</sup> ·h, while the 1.25 g-NMO modified membrane recorded 48.59 L/m<sup>2</sup> ·h. Collectively, the magnitudes of the fluxes in POME filtration were noticeably lower compared to the pure water filtration. This could be due to the presence of contaminants (such as suspended solids, color pigments, organic and inorganic substances) in the POME, which apparently restrict the free flow of permeate through the membranes. Basically, the selectivity and rejection of the contaminants is based on size differences in the pores as well as the surface zeta potential of the membrane [17]. In addition, the results of the pure water flux after POME filtration indicated that membrane with 0.50 g NMO loading had an outstanding recovery with flux of 183.11 L/m<sup>2</sup> ·h.

#### 3.2.2. Rejection of Color Pigment from POME

Figure 7A presents the color removal efficiency for both neat and modified membranes. Noticeably, the membrane with 0.50 g and 0.75 g NMO loading demonstrated outstanding color removal with 74.65% and 72.94%, respectively. It can be noticed that despite the high loading of 1.25 g-NMO, the modified fiber "e" recorded a lower rejection of 47.18% compared to the membrane with 0.50 g-NMO loading. The reason may not be devoid of the inhomogeneity dispersion of the NMO particles which instigated diminution in the effective interacting surface area required to repel the color pigments [36]. Also, Figure 7B display the pictorial visual color differences in feed POME, permeate of the neat and modified membrane as well as pure water. Principally, pore size is a key factor in the separation of color pigment from the POME [17]. This shows that the membrane with pores sizes smaller than the size of the contaminants is capable of giving higher rejection efficiency, though the permeability performance may diminish [50,51]. Besides, the distinct upturn rejection efficiency obtained was not only influenced by the pore sizes but majorly due to the strong negatively charge surface zeta potential. Basically, the surface negativity of the membrane was developed as a result of NMO protonation effect in the presence of water, as explained earlier. In this study, the fed POME has a pH of 8.45 and according to the previous studies, the pH ranging from 8 to 9 was due to the contained color pigments (lignin and tannin) [17,47]. The pigments became negatively charged when the pH of feed was adjusted to acidic conditions (ranging from pH 5.5), this assist in the aggregation of the color pigments [17]. Thus, under this condition of likes negatively charges of color pigments and membrane surface, repulsion of the color pigment prevails during the filtration. This phenomenon considerably improved the color removal efficiency, as observed in the modified fiber (c) and (d).

Figure 8 depicts a simple representation of the color rejection mechanism based on surface negativity of the membrane. As mentioned earlier, the principle of the color rejection was based on the protonation process of both the color pigments present in the diluted POME and NMO to form like charges on the surfaces [47]. During filtration, the hydrated and negatively charged color pigments in the POME migrated towards the surface of the membrane. The presence of water in the medium preceded the protonation of the NMO contained on the surface within the matrix of the membrane, thus resulted in the formation of negative charges. The surging negatively charged pigments were repelled when they approached the membrane surface with like charges [45]. The repulsion of the pigments prevented the deposition of hydrophobic substances (tannin and lignin) on the membrane surface and pore walls. On this basis, the rejection of color pigments not only improves the separation efficiency but also enhance fouling control. Though, some of the un-repelled color pigments with smaller sizes were still able to navigates through the membrane pores along with permeate.

– – **Figure 7.** (**A**) Color rejection performance of neat PVDF-PEG (**a**) and modified Nano-hybrid membrane (**b**–**e**). (**B**). Pictorial view of color differences in the feed POME, permeate of (**a**) neat, (**b**–**e**) modified membranes, and pure water.

**Figure 8.** Schematic of color rejection and antifouling mechanism of modified Nanocomposite NMO-PVDF/PEG membranes during filtration.

#### 3.2.3. Antifouling Performance and Membrane Reusability

Fouling involves deposition of foulants on the surface and pore walls of a polymeric membrane. Several reports have shown that hydrophobic membrane are more prone to severe fouling due to the strong adhesive attraction that exist between the interacting interface [41,52]. This implies that suppressing the hydrophobic properties through the incorporation of hydrophilic Nanoparticles (NMO) into the membrane matrix structure has the potential to curtail foulants deposition, hence the improvement in antifouling properties. On this note, 3 cycles of POME filtration analysis were conducted in accordance to Subramaniam et al. [9] procedure. Throughout, the filtration time and color concentration of the feed were maintained at 180 min and 4285 ADMI for each of the cycles. At the end of each filtration cycle, the membranes were only physically cleaned under running tap water for 15 min. Essentially, this experiment gives the basis to examine the membrane reusability and antifouling performance using FR, RF and IF as the indicators.

Figure 9 presents all the 3 filtration cycles' antifouling performance of the spun membranes. From the cycle 1, 0.5 g-NMO modified membrane recorded the highest FR of 92.32%, while the neat membrane had 27.18%. The FR of 1.25 g, 0.75 g and 0.25 g NMO modified membranes were 69.32%, 86.45% and 61.05%, respectively. Furthermore, the fouling resistance performances of the modified membranes were also superior to the neat membrane. The RF% and IF% of the 0.25 g-NMO, 0.5 g-NMO, 0.75 g-NMO and 1.25 g-NMO were (22.36, 38.95), (61.39, 7.68), (57.64, 13.55) and (19.44, 30.68), respectively, as against (8.44, 72.82) for that of the neat membrane. Even at the end of the third filtration cycle 3, the membrane (c) with 0.50 g NMO loading recorded an FR over 90% with good fouling resistance RF% and IF% at 60.64 and 8.77%, respectively. On the contrary, the flux recovery and antifouling resistance of the neat membrane deteriorated significantly with FR%, RF% and IF% of 10.33%, 4.14% and 91.73%, respectively.

**Figure 9.** Fouling characteristics (FR, RF and IF) of membrane fibers during three (3) consecutive filtration cycles using POME as feed.

Overall, the better antifouling performance of the resultant modified membranes was due to the NMO which essentially improved the hydrophilicity alongside with the negatively surface zeta-potential [25]. Particularly, membrane with 0.50 g loading showed a remarkable improved antifouling performance, and only loss of 1.43% FR with relatively steady RF and IF were observed at the end of the third filtration cycles-3 using diluted-POME as feed. This excellent antifouling

−

− −

−

−

performance of the membrane at this loading may be due to the homogenous dispersion of the Nanoparticles, thereby creating superior interacting surface for effective repellence of foulants [36]. More so, apart from the protonation purpose of the NMO in the presence of water to develop a negative charges on the membrane surface, it is also capable of generating free reactive oxygen species (ROS) to inhibit bio-fouling [36]. Thus, preventing the formation of bio-film and cake layer on the membrane surface and pore walls. In addition, Hikku et al. [42] reported that the NMO antifouling activity has a correlation with magnitude of surface area of the Nanoparticle contacting with the microbes (foulants). This implies that evenly dispersed NMO within the membrane matrix presents larger interacting surface area, thus advancing antifouling as well as flux recoverability [25].

#### 3.2.4. FTIR Analysis of used Membranes

Figure 10 compares the FTIR spectra of the fouled neat membrane (red spectrum) and the modified membrane (blue spectrum) after-used. The modified-membrane had a broad absorbance at 3440, 2926–2304 and 725 cm−<sup>1</sup> , and these can be attributed to the stretch disturbance of O–H, C–H and Mg–O, respectively [40]. The first functional group (O–H) often indicate the degree of surface hydrophilicity and antifouling properties [24,40,53], as accounted by the broad bands of O–H (blue spectrum). In addition, the FTIR spectra suggest that Nano-MgO (NMO) was successfully incorporated into the PVDF-PEG matrix structure using phase inversion technique. On the contrary, no similar band observed in the neat membrane spectra (red spectrum). The noticeable bands in the neat membrane spectra appeared at 1663, 1397, 1175, 1067 and 883 cm−<sup>1</sup> which can be attributed to stretch vibrations of amide-I, methyl bonds, CF2, polysaccharides and complex aromatic functional group, respectively [54,55]. It was noticed that the stretch band at 1175 cm−<sup>1</sup> was peculiar in both spectra which resulted due to intrinsic vibration of asymmetric CF<sup>2</sup> of the PVDF-PEG composition [56]. As suggested by the spectra (red spectrum) of the neat membrane, among the most prevailing functional groups the protein (amide-I 1663 cm−<sup>1</sup> ) and polysaccharides (1067 cm−<sup>1</sup> ) were both included. These are major source of EPS and SMP substance that are usually responsible for the initiation and formation of biofilm which eventually degenerates into cake layer [57]. Therefore, this shows that the neat membrane is highly prone to fouling. On the contrary, the deprotonating process of the Nano-MgO (NMO) improves the intensity of the reactive O-H (blue spectrum) in the modified membrane. Thus, generated O–H denaturize the deposited EPS/SMP as well as improving the hydrophilicity of the polymeric membrane [40]. This phenomenon effectively averts formation of the cake layer on the modified membrane, as observed in Figure 10.

**Figure 10.** FTIR spectra of used neat membrane (red spectrum), and the modified membrane (blue spectrum).

#### *3.3. Performance Appraisal with Literatures*

This section presents a concise comparison of this study with previous works that uses MgO Nanoparticle (NMO) to modified polymeric membranes. In this study, NMO was successfully incorporated into the PVDF-PEG matrix using phase inversion technique. This resulted in considerable augmentation of the hydrophilicity and surface negativity of the spun hollow fibers. At 0.50 g NMO loading, a remarkable water flux, and POME permeability flux along with the color rejection of 198.35 L/m<sup>2</sup> ·h, 61.33 L/m<sup>2</sup> ·h and 74.65% were obtained, respectively. Even after the 3 filtration cycle of a total period of 9 h, the flux recovery percentage remains relatively steady at 92.32%. Previously, Parvizian et al. [39] reported that at a varied loading of NMO blended into PVC matrix, the flux improves and the highest flux was obtained at 0.5 wt % dosage. However, a comprehensive fouling analysis and color rejection of the PVC membrane were not considered. Arumugham et al. [58] used NMO to modify sulfonated polyphenyl sulfone (SPPSU) polymeric membrane and applied to treat oily wastewater. The NMO loading was fixed at 25 wt % while the SPPSU and organic solvent were varied. From the results, a good hydrophilic performance (99.00% of flux recovery) was achieved. However, the improvement in the hydrophilicity may not be attributed to incorporated NMO since the dosage was not varied [58]. Based on this appraisal, it can be deduced that this study not only bridge the information gap, but also reported the significant role of the NMO at varied dosage in improving performance of the hybrid PVDF-PEG membrane.

#### **4. Conclusions**

Hybrid PVDF-PEG hollow fibers blended with NMO at various loading of 0–1.25 g were fabricated using phase inversion technique. The increasing NMO loading has demonstrated a significant effect on the morphology, hydrophilicity, permeability and antifouling properties of the resultants composite hollow fibers. The loading at 0.50 g-NMO presented the most auspicious performance with 198.35 and 61.33 L/m<sup>2</sup> ·h of pure water and POME filtration flux, respectively. After a continues 3 filtration cycles of diluted-POME for a total period of 9 h, the 0.50 g-NMO composite membrane recorded the best flux recovery (FR), reversible fouling (RF) and least irreversible fouling (IF) percentages of 90.98%, 61.39% and 7.68%, respectively. The outstanding performance was due to the homogenous distribution and compatibility of the membrane matrix with the NMO particles at 0.50 g loading. Conversely, after the third filtration cycle-3, a significant deterioration in the FR and fouling resistance (RF and IF) were noticed at higher loading of 1.25 g-NMO with a values of 55.98%, 11.44% and 40.38%, respectively. This was due to the excessive NMO loading which resulted in the formation of numerous aggregated particles within the matrix structure. The agglomerated particles skewed the nucleation and protonation process during the phase separation and crystallization, as well as creating weak spots in the resultant membranes. Therefore, it can be deduced that Nano-hybrid PVDF-PEG membrane with 0.50 g-NMO loading presents the best performance. Overall, the modified fibers presented better performances compared to the neat membrane.

#### **5. Patents**

Malaysian Patent No. PI 2019006570, 2019: A Hybrid System and Method for Treating Palm Oil Mill Effluent.

**Author Contributions:** Conceptualization, M.A., and H.C.M.; methodology, M.A., H.C.M. and P.S.G.; software, M.A.; validation, M.A., H.C.M., P.S.G., K.F.Y., and Z.Z.A.; formal analysis, M.A., H.C.M. and P.S.G.; investigation, M.A.; resources, H.C.M., P.S.G., A.I.M.I. and A.F.I.; data curation, M.A.; writing—original draft preparation, M.A.; writing—review and editing, M.A., H.C.M., P.S.G., K.F.Y., Z.Z.A. and A.I.M.I.; visualization, M.A., H.C.M., and P.S.G.; supervision, H.C.M., K.F.Y., Z.Z.A. and A.I.M.I.; project administration, H.C.M., P.S.G. and A.F.I.; funding acquisition, H.C.M. and M.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the Universiti Putra Malaysia (PUTRA Impak: 9530900), and Tertiary Education Trust Fund (TETF/UNIV/ZARIA/ASTD/2017).

**Conflicts of Interest:** The authors declare no conflict of interest, and also 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**


© 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*

### **Kinetic Studies on the Catalytic Degradation of Rhodamine B by Hydrogen Peroxide: E**ff**ect of Surfactant Coated and Non-Coated Iron (III) Oxide Nanoparticles**

#### **Mohd Shaban Ansari <sup>1</sup> , Kashif Raees <sup>1</sup> , Moonis Ali Khan <sup>2</sup> , M.Z.A. Rafiquee 1,\* and Marta Otero 3,\***


Received: 3 September 2020; Accepted: 26 September 2020; Published: 29 September 2020 -

**Abstract:** Iron (III) oxide (Fe3O4) and sodium dodecyl sulfate (SDS) coated iron (III) oxide (SDS@Fe3O4) nanoparticles (NPs) were synthesized by the co-precipitation method for application in the catalytic degradation of Rhodamine B (RB) dye. The synthesized NPs were characterized using X-ray diffractometer (XRD), vibrating sample magnetometer (VSM), scanning electron microscopy (SEM), transmission electron microscopy (TEM), and Fourier transform infra-red (FT-IR) spectroscopy techniques and tested in the removal of RB. A kinetic study on RB degradation by hydrogen peroxide (H2O2) was carried out and the influence of Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> magnetic NPs on the degradation rate was assessed. The activity of magnetic NPs, viz. Fe3O<sup>4</sup> and SDS@Fe3O4, in the degradation of RB was spectrophotometrically studied and found effective in the removal of RB dye from water. The rate of RB degradation was found linearly dependent upon H2O<sup>2</sup> concentration and within 5.0 × 10−<sup>2</sup> to 4.0 × 10−<sup>1</sup> M H2O2, the observed pseudo-first-order kinetic rates (kobs, s−<sup>1</sup> ) for the degradation of RB (10 mg L−<sup>1</sup> ) at pH 3 and temperature 25 ± 2 ◦C were between 0.4 and 1.7 × 10<sup>4</sup> s −1 , while in presence of 0.1% *w*/*v* Fe3O<sup>4</sup> or SDS@Fe3O<sup>4</sup> NPs, kobs were between 1.3 and 2.8 × 10<sup>4</sup> s <sup>−</sup><sup>1</sup> and between 2.6 and 4.8 × 10<sup>4</sup> s −1 , respectively. Furthermore, in presence of Fe3O<sup>4</sup> or SDS@Fe3O4, kobs increased with NPs dosage and showed a peaked pH behavior with a maximum at pH 3. The magnitude of thermodynamic parameters E<sup>a</sup> and ∆H for RB degradation in presence of SDS@Fe3O<sup>4</sup> were 15.63 kJ mol−<sup>1</sup> and 13.01 kJ mol−<sup>1</sup> , respectively, lowest among the used catalysts, confirming its effectiveness during degradation. Furthermore, SDS in the presence of Fe3O<sup>4</sup> NPs and H2O<sup>2</sup> remarkably enhanced the rate of RB degradation.

**Keywords:** magnetite; co-precipitation method; Rhodamine B; sodium dodecyl sulfate; wastewater treatment

#### **1. Introduction**

Mushrooming industrialization and urbanization are primarily responsible for deteriorating the surface and sub-surface water quality, causing hazardous effects on both aquatic organisms and human health. Among water contaminants, dyes and pigments, which are widely discharged from textile, pharmaceutical, paint, rubber, cosmetic, and confectionary industries effluents [1,2], produce unwanted color to water bodies, resulting in intoxication of ecosystems. Rhodamine B (RB) is a synthetic cationic dye, containing a multi-ring aromatic xanthene core planar structure [3]. It is widely used for dyeing and printing applications [4]. The carcinogenic, mutagenic, and toxic effects of RB have been well reported [5–7], evidencing the need of RB contaminated effluents treatment prior to their discharge. Various treatment methodologies, such as reverse osmosis, ion-exchange, precipitation, adsorption, ozone treatment, catalytic reduction, biodegradation, ultrasonic decomposition, coagulation, electrocoagulation, chemical oxidation, and nano-filtration, have been used for the removal of RB and other dyes from water [8–10]. However, high-cost, long process duration, large energy consumption, regeneration difficulties, and pollutants transfer from one phase to another are the major demerits of the aforementioned processes. Thus, advanced oxidation processes (AOPs) are considered comparatively advantageous since they possess favorable decolorizing ability for reactive dyes [11]. Fenton reaction, which is one of the most effective AOPs, has attracted widespread attention. It is operated at acidic pH in the presence of hydrogen peroxide (H2O2) and ferrous ions while yielding hydroxyl radical with powerful oxidation capacity leading to complete decomposition of organic dyes, thus, converting them into non-toxic lower molecular weight products [12]. In this sense, the Fe2+-H2O<sup>2</sup> Fenton system has been widely used for the oxidative removal of RB from water [11–13].

Recently, iron (III) oxide (Fe3O4) nanoparticles (NPs) have been used for removing various dyes and heavy metals from water [14–16]. These NPs are inert, economical, possess unique magnetic properties, and can be easily separated from reaction medium through an external magnetic field [17–19]. Additionally, Fe3O<sup>4</sup> magnetic NPs exhibit a high surface area, depending on the particle size, and show the ability for surface modification. Furthermore, the interaction of Fe3O<sup>4</sup> NPs with H2O<sup>2</sup> generates hydroxyl and peroxyl radicals, which are able to undergo the oxidative degradation of organic pollutants [20–22]. However, bare Fe3O<sup>4</sup> NPs suffer some shortcomings such as agglomeration, limited adsorption ability, and limited working pH range. Coating of Fe3O<sup>4</sup> NPs with surfactants, polymers, silica, starch, polyelectrolytes, etc., render an enhancement in their surface properties and chemical stability, making them suitable for industrial wastewater treatment and catalytic applications [23–29]. Among surfactants used for coating, the anionic sodium dodecyl sulfate (SDS) is known to enhance the ability of NPs to remove pollutants from wastewater, which has been related with the binding and chelating efficiency of its functional groups [30]. Although no studies were found on the specific case of RB, Fe3O<sup>4</sup> NPs modified with SDS have been successfully used for the adsorptive removal of several dyes from water, including tolonium chloride [31], Basic Blue 41 [32], or Brilliant Green [33].

The present work was undertaken with the aim of developing an efficient, eco-friendly, and economical treatment for the removal of cationic dyes from water. For this purpose, Fe3O<sup>4</sup> NPs were synthesized, coated with SDS, and tested as catalyst for the degradation of RB under the presence of H2O2. The synthesized Fe3O<sup>4</sup> and SDS-coated Fe3O<sup>4</sup> (SDS@Fe3O4) NPs were thoroughly characterized through XRD, VSM, SEM, TEM, and FT-IR techniques. The main novelty of this work was the comparative study of the dye degradation by H2O<sup>2</sup> under three different situations, namely, in absence of ferrous NPs, in the presence of Fe3O<sup>4</sup> NPs, and in the presence of SDS@Fe3O<sup>4</sup> NPs. Kinetic experiments were carried out to explore the influence of these catalysts dosages, H2O2, SDS, and solution pH on the RB degradation rate.

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

#### *2.1. Chemicals and Reagents*

Rhodamine B (RB: AR grade 80%; CDH, New Delhi, India), hydrochloric acid (HCl: AR grade 36%; Fisher Scientific, Mumbai, India), hydrogen peroxide (H2O2: 35% *v*/*v*, Merck, Mumbai, India), sodium dodecyl sulphate (SDS: 99%; CDH, New Delhi, India), Ammonia solution (NH4OH: 25% with purity index 99%, Thermo Fisher Scientific, Mumbai, India), ferrous chloride dihydrate (FeCl2. 2 H2O: 99%; CDH, New Delhi), ferric Chloride (FeCl3: 97.0%; CDH, New Delhi, India), and sodium hydroxide

pellets (NaOH: 97%, Merck, Mumbai, India) were used as supplied. All the other reagents used during the experimental work were of reagent grade. All the solutions were prepared in deionized (DI) water. The stock solutions of NaOH (1.0 M) and SDS (1.0 × 10−<sup>2</sup> M) were prepared in DI water. The stock solution (500 mgL−<sup>1</sup> ) of dye was prepared by dissolving 50 mg RB in 100 mL DI water. Likewise, 250 mL stock solution of H2O<sup>2</sup> was prepared by dissolving 25 mL of H2O<sup>2</sup> in DI water. The stock solution of HCl (0.1 M) was prepared in 100 mL DI water.

#### *2.2. Synthesis and Surfactant Coating of Fe3O<sup>4</sup> Magnetic NPs*

Magnetic nanoparticles were synthesized by adopting the co-precipitation method as described in the literature [34]. Briefly, Fe3O<sup>4</sup> NPs were synthesized by mixing 20.0 g of FeCl<sup>3</sup> (0.4 M) and 10.0 g of FeCl2.2H2O (0.2 M) into a 1.0 L conical flask. These iron salts were dissolved in 300 mL DI water. The mixture was purged with N<sup>2</sup> gas and stirred for about an hour. Then, liquor ammonia (25%) was added drop-wise in the flask. The pH of the solution in flask was further increased to ~10 by adding 2.0 M of NaOH solution. The temperature of the solution was then raised to 70 ◦C with stirring and purging of N<sup>2</sup> gas for 5 h. Black precipitate was formed in the flask. It was filtered, washed with acetone, and thereafter with DI water to a neutral pH value. The precipitate was then dried at 70 ◦C in a vacuum oven. The synthesis of Fe3O<sup>4</sup> NPs can be given by the following reaction:

$$\mathrm{Fe}^{2+} + 2\mathrm{Fe}^{3+} + 8\mathrm{OH}^- \rightarrow \mathrm{Fe\_3O\_4} + 4\mathrm{H\_2O}$$

To prepare the SDS-coated Fe3O<sup>4</sup> NPs, FeCl3.6H2O (20 g, 0.40 M), FeCl2.4H2O (10 g, 0.20 M), and SDS (8.64 g, 0.10 M) were taken into the conical flask of 1.0 L capacity containing 300 mL DI water. The overhead stirrer was used to mix the reactants properly. The solution was stirred vigorously for 45 min under the N<sup>2</sup> gas atmosphere. Then, 200 mL of 25% ammonium hydroxide solution was added drop-wise into the above solution until the pH of the resulting solution reached 9–11. The pH of the reaction medium was further raised to 14 by adding 2.0 M NaOH solution drop-wise. The mixture was then stirred vigorously under N<sup>2</sup> gas purging for 5 h. The black precipitate that formed was filtered and washed with acetone and DI water until the pH came to a neutral value.

#### *2.3. Characterization*

The crystallinity and phase composition of Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> NPs were studied by X-ray diffraction (XRD: MiniFlex II, Rigaku, Tokyo, Japan) analysis equipped with a Cu Kα radiation source (with λ = 1.5406 nm). The surface functionalities present over Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> NPs surface were determined by Fourier infra-red spectrometer (FT-IR: Nicolet iS50, Thermo Fisher Scientific, Madison, WI, USA). The surface morphology and particle size were analyzed by scanning electron microscopy (SEM: JSM-5600LV, JEOL, Tokyo, Japan) and transmission electron microscopy (TEM: CM120, Philips, Amsterdam, The Netherlands). The magnetic properties of Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> NPs were determined using a vibrating sample magnetometer (VSM: 7307, Lakeshore, Westerville, OH, USA).

#### *2.4. Degradation Kinetic Experiments*

A Genesys 10S UV–visible spectrophotometer (Thermo Fisher Scientific, Madison, WI, USA) was used to monitor the change in the absorbance intensity of RB during its degradation under the varying reaction conditions. The spectrophotometer was provided with multiple cell holders in which a 3.0 mL quartz cuvette with a path length of 10 mm was used to measure absorbance. All the kinetic experiments were performed at a constant temperature of 25.0 ± 0.2 ◦C by using a thermostatic water-bath. A 0.1% *w*/*v* of magnetic Fe3O<sup>4</sup> NPs was put together with RB solution with an initial concentration of 10 mg L−<sup>1</sup> into a three necked round bottom flask of 100 mL capacity. Solution pH was adjusted by adding hydrochloric acid or sodium hydroxide solution and monitored by using a pH meter. The reaction vessel containing RB solution and magnetic Fe3O<sup>4</sup> NPs was kept in the

water-bath to equilibrate with the required temperature. The reaction was started with the addition of 5.0 × 10 −2 to 4.0 × 10 <sup>−</sup><sup>1</sup> M H2O<sup>2</sup> and zero time was taken when the half of the amount of H2O<sup>2</sup> was added. The concentration of RB was spectrophotometrically analyzed at its maximum absorbance wavelength (λmax: 554 nm) at constant time intervals. All the kinetic experiments were carried out under pseudo-first-order conditions in which H2O<sup>2</sup> was kept in excess over RB. The progress of the reaction gradually resulted in the decrease of RB concentration and the values of the pseudo-first-order rate constants were obtained from the slopes of the plots of ln (absorbance) versus time. Each kinetic run was carried out in triplicate to check their repeatability and the rate constant was observed to be within the error limits of ~5%.

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

#### *3.1. Characterization of Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> NPs*

#### 3.1.1. X-ray Diffraction (XRD)

Figure 1A shows the XRD patterns obtained for the synthesized Fe3O<sup>4</sup> NPs and it confirms the nanocrystal structure and phase purity of Fe3O<sup>4</sup> NPs. The diffraction peaks appeared at 2θ = 30.26 ◦ , 35.5◦ , 43.12 ◦ , 53.74 ◦ , 57.10 ◦ , and 62.92 ◦ corresponding to planes (220), (311), (400), (422), (511), and (440), respectively [35], consistent with standard magnetite database (JCPDS-19-0629), indicating a highly crystalline nature of Fe3O<sup>4</sup> NPs. Figure 1B shows the XRD patterns for SDS@Fe3O<sup>4</sup> NPs with reduced peak intensity due to the SDS coating over Fe3O<sup>4</sup> surface. This confirms crystalline-to-amorphous transition of Fe3O<sup>4</sup> NPs due to SDS coating during SDS@Fe3O<sup>4</sup> NPs synthesis [36].

**Figure 1.** X-ray diffraction patterns for synthesized iron (III) oxide (Fe3O<sup>4</sup> ) (**A**) and sodium dodecyl sulfate (SDS) coated iron (III) oxide (SDS@Fe3O<sup>4</sup> ) (**B**).

#### 3.1.2. Fourier Transform Infrared Spectroscopy (FTIR)

The FT-IR spectra of Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> NPs are shown in Figure 2.

The two peaks at 585 and 435 cm−<sup>1</sup> , as shown in Figure 2A, correspond to the Fe-O bond vibrations of Fe3O<sup>4</sup> NPs [37]. From these observations, it is confirmed the spinel structure of Fe3O<sup>4</sup> NPs and also inferred the existence of the difference in the bond length in Fe-O. The peak at 3424 cm−<sup>1</sup> in Figure 2A was associated to the O-H stretching vibrations arising from the hydroxyl group due to the presence of water molecules associated with Fe3O<sup>4</sup> [38]. The H-O-H bending of water molecules in Figure 2A is observed at 1631 cm−<sup>1</sup> in Fe3O<sup>4</sup> NPs [39]. The FTIR spectrum of SDS@Fe3O<sup>4</sup> NPs is shown in Figure 2B, which displayed a new absorption peak at 1252 cm−<sup>1</sup> due to the stretching vibration of S=O groups of SDS and the presence of peaks at 2929 cm−<sup>1</sup> and 2842 cm−<sup>1</sup> , which were assigned to the stretching mode for aliphatic C-H groups of SDS [40]. The peak at 1635 cm−<sup>1</sup> in SDS@Fe3O<sup>4</sup> (Figure 2B) was

attributed to the H-O-H bending of water molecules and that at 3431 cm−<sup>1</sup> was due to stretching vibration of hydroxyl group on the surface of the NPs. The presence of two peaks at 547 cm−<sup>1</sup> and at 474 cm−<sup>1</sup> in Figure 2B is attributed to Fe-O bonds in SDS-modified Fe3O<sup>4</sup> [41]. Thus, the FTIR results confirmed successful synthesis of Fe3O<sup>4</sup> NPs and their surface modifications through the adsorption of SDS molecules.

**Figure 2.** FTIR spectra of Fe3O<sup>4</sup> (**A**) and SDS@Fe3O<sup>4</sup> (**B**).

#### 3.1.3. Scanning Electron Microscopy (SEM)

The SEM micrograph of the synthesized magnetite (Fe3O4) NPs is shown in Figure 3A. It can be observed that the NPs exhibit spherical surface morphology, having a particle size lower than 100 nm scale with low polydispersity. The SEM image of the SDS@Fe3O<sup>4</sup> NPs is shown in Figure 3D on the scale of up to 5 µm. The image depicts successful functionalization of Fe3O<sup>4</sup> by SDS and the larger dispersion of SDS@Fe3O<sup>4</sup> as compared with Fe3O<sup>4</sup> NPs.

**Figure 3.** SEM image of Fe3O<sup>4</sup> (**A**); TEM image of Fe3O<sup>4</sup> (**B**); histogram images of Fe3O<sup>4</sup> (**C**); SEM image of SDS@Fe3O<sup>4</sup> (**D**); TEM image of SDS@Fe3O<sup>4</sup> (**E**); histogram images of SDS@Fe3O<sup>4</sup> (**F**). Note that scales of figures A and B are different.

#### 3.1.4. Transmission Electron Microscopy (TEM)

The TEM micrograph of pristine Fe3O<sup>4</sup> NPs (Figure 3B) on the scale of up to 20 µm shows their spherical shape with a narrow range particle size distribution centered at 9 ± 2 nm, as demonstrated by the histogram in Figure 3C. The TEM image of the SDS@Fe3O<sup>4</sup> NPs is illustrated in Figure 3E. After coating with SDS, the size of SDS@Fe3O<sup>4</sup> NPs appears to be smaller, as shown by the histogram (Figure 3F). This might be due to the coating of Fe3O<sup>4</sup> NPs with SDS, which hinders NPs agglomeration.

#### 3.1.5. Vibrating Sample Magnetometer (VSM)

The magnetic behavior of Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> NPs was studied by using VSM. As it may be seen in Figure 4, both Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> showed superparamagnetic behavior with different magnetic saturations level. The specific magnetic saturation magnitudes for Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> NPs were 60.0 and 50.0 emug −1 , respectively, as displayed in Figure 4. Comparatively lower magnetic saturation of SDS@Fe3O<sup>4</sup> NPs might be due to their coating with SDS [42]. In order to avoid aggregation of Fe3O<sup>4</sup> NPs, which may severely reduce their catalytic efficiency, coating with SDS was executed in this work. In any case, as for the large magnetic saturation and superparamagnetic property of SDS@Fe3O<sup>4</sup> NPs (Figure 4), such a coating did not affect the high efficiency in magnetic separation and recovery.

**Figure 4.** Magnetization curve for Fe3O<sup>4</sup> and for SDS@Fe3O<sup>4</sup> at room temperature.

#### *3.2. Degradation of RB by H2O<sup>2</sup>*

The repetitive scans of the reactant mixture containing RB (10 mg L −1 ) and H2O<sup>2</sup> (2 × 10 <sup>−</sup><sup>1</sup> M) were recorded at constant time intervals of ten minutes in the visible region (460–600 nm). The temperature and pH were kept constant at 25 ± 0.2 ◦C and 3, respectively. These spectra, which are shown in Figure 5A, indicated that the absorbance intensities at λmax (554 nm) progressively decreased with time. A decrease in the absorbance intensities was due to the degradation of RB by H2O2.

The degradation of RB can be represented by the following representative reaction and rate Equation (1):

$$\text{RB} + \text{H}\_2\text{O}\_2 \rightarrow \text{Oxidized products} + \text{H}\_2\text{O}, \text{Rate} = \frac{-\text{d}[\text{RB}]}{\text{dt}} = \text{k}\_{\text{obs}}[\text{RB}], \tag{1}$$

where kobs is the observed value of the rate constant and was calculated from the slope of the plot of ln [RB] 0 [RB] t versus t. The terms [RB] 0 and [RB] t are the concentrations of RB at time zero and at any time t, respectively. The observed rate constant depends upon the concentration of H2O<sup>2</sup> as given by Equation (2). The order of the reaction is assumed to be x with respect to the concentration of H2O2.

$$\mathbf{k}\_{\rm obs} = \mathbf{k} \left[ \mathbf{H}\_2 \mathbf{O}\_2 \right]^\chi \tag{2}$$

where k is the specific rate constant with respect to H2O<sup>2</sup> concentration. The values of k and x were respectively obtained from the intercept and slope of the plot of log kobs versus log [H2O2].

**Figure 5.** UV–visible spectra of Rhodamine B (RB) at various degradation times. In absence of Fe3O<sup>4</sup> (**A**); in presence of Fe3O<sup>4</sup> (**B**); and in presence of SDS@Fe3O<sup>4</sup> (**C**). (Reaction conditions: 10 mg L <sup>−</sup><sup>1</sup> RB, 2.0 × 10 <sup>−</sup><sup>1</sup> M H2O<sup>2</sup> , 0.1% *w*/*v* Fe3O<sup>4</sup> NPs, 0.1% *w*/*v* SDS@Fe3O<sup>4</sup> , pH 3, and temperature 25 ± 2 ◦C).

The rate of RB degradation was studied at varied concentrations of H2O<sup>2</sup> in the range from 5.0 × 10 −2 to 4.0 × 10 <sup>−</sup><sup>1</sup> M while keeping a RB concentration of 10 mg L <sup>−</sup><sup>1</sup> at pH 3 and temperature 25 ± 0.2 ◦C. The values of rate constants were calculated and the plot of rate constant versus H2O<sup>2</sup> concentration (Figure 6) shows a linear dependence of the rate constant values on H2O<sup>2</sup> concentration.

**Figure 6.** Plots of kobs versus hydrogen peroxide (H2O<sup>2</sup> ) concentration for the degradation of RB. (Reaction conditions: 10 mg L <sup>−</sup><sup>1</sup> RB, 5.0 × 10 −2 to 4.0 × 10 <sup>−</sup><sup>1</sup> M H2O<sup>2</sup> , pH 3, and temperature 25 ± 2 ◦C).

#### *3.3. Degradation of RB in the Presence of Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> NPs*

The addition of 0.1% *w*/*v* of Fe3O<sup>4</sup> NPs to the solution containing RB and H2O<sup>2</sup> increased the rate of degradation of RB, as is evident from the decrease in the rate of absorbance intensities with time, which is presented in Figure 5). The increase in the degradation rate of RB can be attributed to the catalytic role of Fe3O<sup>4</sup> NPs. The degradation rate was further increased in the presence of 0.1% *w*/*v* SDS@Fe3O<sup>4</sup> NPs as displayed in Figure 5C.

In order to assess the effect of pH, the degradation rate of RB was studied in the pH range 1–10 by adjusting it with HCl/NaOH solutions. The observed results are presented in Figure 7. The plot of the rate constant versus pH (Figure 7) demonstrates that the values of the rate constant increase with pH until pH 3. Thereafter, on further increasing the pH beyond 3, the values of the rate constant decreased. Thus, a peaked behavior plot was obtained with the maximum degradation rate at pH 3.

**Figure 7.** Effect of pH on the Rhodamine B (RB) degradation process. In presence of Fe3O<sup>4</sup> (**A**); and in presence of SDS@Fe3O<sup>4</sup> (**B**). (Reaction conditions: 10 mg L <sup>−</sup><sup>1</sup> RB, 2.0 × 10 <sup>−</sup><sup>1</sup> M H2O<sup>2</sup> , 0.1% *w*/*v* Fe3O<sup>4</sup> , 0.1% *w*/*v* SDS@Fe3O<sup>4</sup> , and temperature 25 ± 2 ◦C).

The influence of the magnetic Fe3O<sup>4</sup> NPs dosage on the RB degradation rate was studied in the range between 0.02% and 0.2% *w*/*v* Fe3O4. The respective concentrations of H2O<sup>2</sup> and RB were set at 2.0 × 10 <sup>−</sup><sup>1</sup> M and 10 mg L −1 , while the pH and temperature of the solution were 3 and 25 ± 0.2 ◦C, respectively. The increase in the amount of Fe3O<sup>4</sup> increased the RB degradation rate, as shown by data graphically presented in Figure 8A. Furthermore, as it may be seen in Figure 8B, the influence of SDS@Fe3O<sup>4</sup> concentration on the RB degradation rate showed the same pattern observed for Fe3O4, but with higher values of the rate constant.

**Figure 8.** Plots of kobs versus varying concentration of magnetic nanoparticles (NPs) for the degradation of the RB in presence of Fe3O<sup>4</sup> (**A**), and in presence of SDS@Fe3O<sup>4</sup> (**B**). (Reaction conditions: 10 mg L −1 RB, 2.0 × 10 <sup>−</sup><sup>1</sup> M H2O<sup>2</sup> , 0.1% *w*/*v* Fe3O<sup>4</sup> , 0.1% *w*/*v* SDS@Fe3O<sup>4</sup> , pH 3, and temperature 25 ± 2 ◦C).

The observed enhancement in the rate of the RB degradation in presence of Fe3O<sup>4</sup> can be described through the production of highly reactive hydroxyl radicals due to the interaction between the NPs and H2O<sup>2</sup> [43,44], followed by the formation of peroxyl radicals and the subsequent oxidation of RB by these radicals, as described by the following reactions:

(i) Fe2<sup>+</sup> + H2O<sup>2</sup> → Fe3<sup>+</sup> + HO<sup>∗</sup> + OH<sup>−</sup> The possible reactions of free radicals are:

(ii) Fe2<sup>+</sup> + HO<sup>∗</sup> → Fe3<sup>+</sup> + OH<sup>−</sup> ,


(v) HO<sup>∗</sup> + HO<sup>∗</sup> → H2O2,

(vi) HO<sup>∗</sup> + RB → Products,

(vii) HO<sup>∗</sup> 2 + RB → Products.

Thus, the oxidation of RB by HO\*and HO<sup>2</sup> ∗ radicals leads to a decrease in its concentration. As for the RB degradation in the presence and absence of Fe3O<sup>4</sup> and SDS@Fe3O4, results shown in Figure 5 to 8 allow to state that RB degradation was enlarged under the presence of Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> NPs. As for the degradation rate of RB, it was linearly dependent on the initial concentration of H2O<sup>2</sup> in the absence of NPs. The RB degradation was comparatively higher in presence of SDS@Fe3O<sup>4</sup> than Fe3O<sup>4</sup> and increased with the increase in the dosage of either Fe3O<sup>4</sup> or SDS@Fe3O<sup>4</sup> NPs. The variations in pH displayed a similar influence on the RB degradation rate in the presence of Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> NPs, the rate showing a peaked behavior. From these observations, it was confirmed that the reaction proceeded through the formation of highly reactive free radicals in the presence of Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> NPs due to the interaction between Fe2<sup>+</sup> and H2O2, as described by reactions (i) to (vii). The increase in the amount of the NPs increases the production of HO<sup>∗</sup> radicals (step (i)) and, therefore, an enhancement in the RB degradation rate was observed with the increase in the NPs dosage, which is coincident with previous observations [45]. As shown in Figure 6, the RB degradation rate increased from 0.4 to 1.7 × 10<sup>4</sup> s <sup>−</sup><sup>1</sup> with increasing H2O<sup>2</sup> concentration in the absence of NPs. In Figure 9, under the presence of NPs, larger degradation rates are represented, varying between 1.3 and 2.8 × 10<sup>4</sup> s −1 in the case of Fe3O<sup>4</sup> and between 2.6 and 4.8 × 10<sup>4</sup> s −1 in the case of SDS@Fe3O4, which compare rather well with published rate constants for the catalytic degradation of RB (Table S1 in the Supplementary Materials). However, as it may be seen in Figure 9, in the presence of Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> NPs, the rate constant value increased with the concentration of H2O<sup>2</sup> until it was 2.5 × 10−<sup>1</sup> M H2O2, and thereafter decreased with H2O<sup>2</sup> concentration. After the maximum, this decreasing effect in the rate constant with the increase in the H2O<sup>2</sup> concentration was due to the other free radical reactions taking place in steps (ii) to (v). Thus, at higher concentrations of H2O2, the side reactions scavenged the HO<sup>∗</sup> radicals and decreased the concentration of free radicals available to oxidize the dye and, therefore, the rate of the reaction decreased [46].

The rate of degradation of RB was highly pH-dependent and, as it is shown in Figure 7, the maximum rate of degradation was observed at pH 3 in the presence and the absence of NPs. At high concentrations of H<sup>+</sup> ions (pH < 3), peroxide gets solvated to form stable oxonium ions, which enhanced the activity of H2O<sup>2</sup> and restricted the generation of hydroxyl radicals [47–49]. Moreover, the excess of H<sup>+</sup> ions acts as hydroxyl radical scavenger and, with the increase in H<sup>+</sup> ions, the concentration of HO<sup>∗</sup> radicals decreases, thus, decreasing the rate of reaction [48]. Furthermore, the strong electrostatic interaction between the anionic surfactant head groups and cationic dye molecules at lower pH also decreases the rate of RB degradation. The observed lower rate of reaction at higher pH may be related to the formation of the Fe3+-complexes, which decreases the dissolved Fe2<sup>+</sup> ions that were available to generate free radicals [49].

**Figure 9.** Effect of H2O<sup>2</sup> concentration on RB degradation. In presence of Fe3O<sup>4</sup> (**A**); and in presence of SDS@Fe3O<sup>4</sup> **(B**). (Reaction conditions: 10 mg L <sup>−</sup><sup>1</sup> RB, 5.0 × 10 −2 to 4.0 × 10 <sup>−</sup><sup>1</sup> M H2O<sup>2</sup> , 0.1% *w*/*v* Fe3O<sup>4</sup> , 0.1% *w*/*v* SDS@Fe3O<sup>4</sup> , pH 3, and temperature 25 ± 2 ◦C).

The higher degradation rate of RB in the presence of SDS@Fe3O<sup>4</sup> NPs in comparison with bare Fe3O<sup>4</sup> NPs that is observed in Figure 8, might be due to the larger capture of RB by SDS@Fe3O<sup>4</sup> than by Fe3O4. Thus, the generated free radicals at the NPs surface can readily attack the attached RB and thus leading to the increase in RB degradation rate. Binding of RB to the SDS@Fe3O<sup>4</sup> surface can be explained by the electrostatic interaction between the anionic surfactant and protonated cationic dye at pH 3 [50].

#### *3.4. E*ff*ect of SDS Concentration and Fe3O<sup>4</sup> NPs Dosage on RB Degradation*

The addition of SDS at varied concentrations (5.0 × 10 -4 to 5.0 × 10 <sup>−</sup><sup>2</sup> M) to a solution containing RB (10 mgL −1 ), H2O<sup>2</sup> (2.0 × 10 <sup>−</sup><sup>1</sup> M), and Fe3O<sup>4</sup> (0.1% *w*/*v*) NPs at pH 3 resulted in an increase in the rate of the degradation reaction, as shown in Figure 10. On the other hand, an increase in the amount of Fe3O<sup>4</sup> NPs from 0.02% to 0.2% *w*/*v* at a fixed concentration of SDS (2.0 × 10 <sup>−</sup><sup>2</sup> M) also increased the rate of RB degradation, as shown in Figure 11.

**Figure 10.** Effect of sodium dodecyl sulfate (SDS) concentration on RB degradation. (Reaction conditions: 10 mg L <sup>−</sup><sup>1</sup> RB, 2.0 × 10 <sup>−</sup><sup>1</sup> M H2O<sup>2</sup> , 0.1% *w*/*v* Fe3O<sup>4</sup> , pH 3, and temperature 25 ± 2 ◦C).

The respective degradation rates of RB in presence of SDS and Fe3O<sup>4</sup> can be represented by Equations (3) and (4).

$$\text{RB} + \text{D}\_{\text{n}} \stackrel{\text{K}\_{\text{S}}}{\Leftrightarrow} \text{RB}\_{\text{mic}} \stackrel{\text{H}\_{2}\text{O}\_{2}}{\rightarrow} \text{Products}\_{\text{A}} \tag{3}$$

$$\text{Fe}\_3\text{O}\_4 + \text{RB}\_{\text{mic}} \overset{\text{K}\_3}{\Leftrightarrow} \text{RB}\_{\text{mic}} - \text{Fe}\_3\text{O}\_4 \overset{\text{H}\_2\text{O}\_2}{\leftrightarrow} \text{Products.} \tag{4}$$

The presence of SDS micelles (Dn) partitions RB into micellar (RBmic ) and aqueous pseudo-phases resulting into the retardation of RB oxidation with H2O<sup>2</sup> in the presence of SDS, which may be related to the electrostatic repulsion and, therefore, separation between the species involved in the reaction. However, in the presence of Fe3O4, micellised RB (RBmic ) is incorporated to the NPs surface to form RBmic−Fe3O<sup>4</sup> where H2O<sup>2</sup> interacts to form reactive HO ∗ radicals readily available to oxidize RB at the same site. Therefore, RB degradation is catalyzed and the rate of the reaction increases with increasing SDS concentration in the presence of Fe3O<sup>4</sup> NPs (Figure 10) and also with increasing the amount of Fe3O<sup>4</sup> NPs in the presence of SDS (Figure 11). In Figure 10, a two steps increase of the degradation may be observed, which may be related to the formation of premicellar aggregates below the critical micelle concentration (cmc) of SDS and micelles above cmc [8,10,51,52], then increasing micelles formation with SDS concentration. Regarding Figure 11, at a SDS concentration above cmc, an increasing degradation rate occurred under increasing Fe3O<sup>4</sup> concentration, as previously observed in Figure 8 and explained by reactions (i) to (vii). These results are in agreement with previous studies on RB photocatalytic degradation [53,54].

**Figure 11.** Effect of Fe3O<sup>4</sup> concentration on RB degradation. (Reaction conditions: 10 mg L <sup>−</sup><sup>1</sup> RB, 0.02% to 0.2% *w*/*v* Fe3O<sup>4</sup> , 2.0 × 10 <sup>−</sup><sup>2</sup> M SDS, pH 3, and temperature 25 ± 2 ◦C).

#### *3.5. E*ff*ect of Temperature on RB Degradation*

The effect of temperature on RB degradation (10 mg L −1 ) in aqueous solutions in the presence of H2O<sup>2</sup> (2.5 × 10 <sup>−</sup><sup>1</sup> M) and at pH 3 was studied at varied temperatures ranging from 25 to 60 ◦C (because above 60 ◦C, due to thermal disintegration of H2O<sup>2</sup> and free radicals, the rate of RB degradation slowed down) in the absence or in the presence of Fe3O<sup>4</sup> NPs (0.1% *w*/*v*), Fe3O<sup>4</sup> NPs (0.1% *w*/*v*) together with SDS (2.0 × 10 <sup>−</sup><sup>2</sup> M) or SDS@Fe3O<sup>4</sup> (0.1% *w*/*v*).

The energy of activation was calculated using the Arrhenius equation Equation (5), which gave a straight line plot for log k versus 1/T.

$$\log \mathbf{k}\_{\rm obs} = -\frac{\mathbf{E}\_{\rm a}}{2.303 \,\mathrm{RT}} + \log \mathbf{A}\_{\rm o} \tag{5}$$

where Ea is the activation energy (kJ mol −1 ), R (8.314 J mol <sup>−</sup>1K −1 ) is the universal gas constant, T is the temperature in Kelvin (K), A<sup>o</sup> is the frequency factor, and kobs is the measured first-order rate constant. The E<sup>a</sup> was determined from the slope and values are given in Table 1.

The value of ∆H (enthalpy of activation) and ∆S (entropy of activation) were calculated using the Erying equation Equation (6).

$$\ln\left(\frac{\mathbf{k}\_{\rm obs}}{\mathbf{T}}\right) = -\frac{\Delta \mathbf{H}}{\mathbf{R}} \times \frac{1}{\mathbf{T}} + \ln\frac{\mathbf{k}\_{\rm B}}{\mathbf{h}} + \frac{\Delta \mathbf{S}}{\mathbf{R}}\tag{6}$$

where k<sup>B</sup> is the Bolzmann's constant and h is the Plank's constant. A plot of ln (kobs/T) versus 1/T produces a straight line and the values of ∆H and ∆S may be obtained from the slope and the intercept, respectively. The so determined ∆H and ∆S values are given in Table 1.

As it may be seen in Table 1, good fittings (R<sup>2</sup> > 0.94) to the Erying equation were obtained within the temperature range here considered. The largest E<sup>a</sup> and ∆H determined for RB degradation were those in the absence of NPs. These values progressively decreased in the presence of Fe3O<sup>4</sup> NPs, Fe3O<sup>4</sup> NPs together with SDS and SDS@Fe3O4, which provided the lowest E<sup>a</sup> and ∆H. Regarding the ∆S, although the effect was not so remarkable as for E<sup>a</sup> and ∆H, slightly lower values were also determined under the presence of NPs. These results point to the energetically favorable effect of Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> NPs, which confirms that these are efficient catalysts.

**Table 1.** Activation parameters determined for RB degradation by H2O<sup>2</sup> in the absence and presence of Fe3O<sup>4</sup> and SDS and SDS@Fe3O<sup>4</sup> .


Reaction conditions: 10 mg L−<sup>1</sup> RB, 2.0 × 10−<sup>1</sup> M H2O2, 0.1% *w*/*v* Fe3O<sup>4</sup> (when present), 0.1% *w*/*v* SDS@Fe3O<sup>4</sup> (when present), 2.0 × 10−<sup>2</sup> M SDS, pH 3, and varied temperatures (between 25 and 60 ± 2 ◦C).

#### **4. Conclusions**

In this work, Fe3O<sup>4</sup> NPs were synthesized, coated with SDS to synthesize SDS@Fe3O<sup>4</sup> NPs, and both tested as catalysts for the oxidation of RB under H2O2. The main novelty was to compare the dye degradation under three different situations, namely, in presence of just H2O2, of H2O<sup>2</sup> and Fe3O<sup>4</sup> NPs, and H2O<sup>2</sup> and SDS@Fe3O<sup>4</sup> NPs. Observed pseudo-first-order kinetic rates (kobs, s−<sup>1</sup> ) for the degradation of RB (10 mg L−<sup>1</sup> ) at pH 3 and temperature 25 ± 2 ◦C were between 0.4 and 1.7 × 10<sup>4</sup> s −1 , linearly dependent upon H2O<sup>2</sup> concentrations within 5.0 × 10−<sup>2</sup> to 4.0 × 10−<sup>1</sup> M. Under identical experimental conditions, except for the presence of 0.1% *w*/*v* NPs, the observed rates increased to values between 1.3 and 2.8 × 10<sup>4</sup> s −1 in the case of Fe3O<sup>4</sup> and between 2.6 and 4.8 × 10<sup>4</sup> s −1 in the case of SDS@Fe3O4. Fe3O<sup>4</sup> NPs with H2O<sup>2</sup> gave readily the highly reactive hydroxyl radicals, which enhanced the rate of RB degradation. Furthermore, an increased catalytic effect was observed for SDS@Fe3O<sup>4</sup> because the SDS coating avoided Fe3O<sup>4</sup> aggregation and the consequent efficiency depletion. However, under the presence of Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> NPs, kobs did not increase linearly with H2O<sup>2</sup> concentration but just until 2.5 × 10−<sup>1</sup> M H2O2, then decreased with increasing H2O<sup>2</sup> concentration, which was associated to free radical competitive reactions. On the other hand, it was verified that the addition of SDS molecules to the dye solution containing Fe3O<sup>4</sup> also increased the rate of reaction, which was related to the incorporation of micellized RB ions onto the Fe3O<sup>4</sup> NPs surface. Overall, this work demonstrated that the application of Fe3O<sup>4</sup> and SDS@Fe3O<sup>4</sup> along with H2O<sup>2</sup> can be an efficient method for the rapid removal of cationic dyes from wastewater in line with the green chemistry principles.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4360/12/10/2246/s1, Table S1: Published results on the rate constant (kobs, s−<sup>1</sup> ) for the catalytic degradation of RB using different catalysts.

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

**Funding:** Marta Otero is thankful to the Portuguese "Fundação para a Ciência e a Tecnologia" (FCT) for the Investigator Program (IF/00314/2015). We would also like to thank FCT/Ministério da Ciência, Tecnologia e Ensino Superior (MCTES) for the financial support to CESAM (UIDP/50017/2020+UIDB/50017/2020) through national funds.

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


© 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* **Molecularly Imprinted Polymers for the Removal of Antide-Pressants from Contaminated Wastewater**

**Tjasa Gornik 1,2, Sudhirkumar Shinde 3,4 , Lea Lamovsek <sup>5</sup> , Maja Koblar 2,6, Ester Heath 1,2, Börje Sellergren <sup>3</sup> and Tina Kosjek 1,2,\***


**Abstract:** Selective serotonin reuptake inhibitors (SSRIs) are a class of antidepressants regularly detected in the environment. This indicates that the existing wastewater treatment techniques are not successfully removing them beforehand. This study investigated the potential of molecularly imprinted polymers (MIPs) to serve as sorbents for removal of SSRIs in water treatment. Sertraline was chosen as the template for imprinting. We optimized the composition of MIPs in order to obtain materials with highest capacity, affinity, and selectivity for sertraline. We report the maximum capacity of MIP for sertraline in water at 72.6 mg g−<sup>1</sup> , and the maximum imprinting factor at 3.7. The MIPs were cross-reactive towards other SSRIs and the metabolite norsertraline. They showed a stable performance in wastewater-relevant pH range between 6 and 8, and were reusable after a short washing cycle. Despite having a smaller surface area between 27.4 and 193.8 m<sup>2</sup> ·g −1 , as compared to that of the activated carbon at 1400 m<sup>2</sup> ·g −1 , their sorption capabilities in wastewaters were generally superior. The MIPs with higher surface area and pore volume that formed more non-specific interactions with the targets considerably contributed to the overall removal efficiency, which made them better suited for use in wastewater treatment.

**Keywords:** molecular imprinting; polymer; wastewater treatment; sertraline; cross-reactivity; SSRI; template; sorbent

#### **1. Introduction**

The fast population growth, advances in industry, and increased agricultural activity have greatly influenced the environment. In order to continue with the current pace, we need solutions in environmental management, especially wastewater (WW) reuse. The development in the area of sample preparation and instrumentation has put the removal of trace-level emerging contaminants in the forefront of environmental research [1,2]. Among them, pharmaceuticals are a very problematic group, since they are particularly designed to have a pharmacological effect on humans or animals, thus potentially yielding adverse effects in living organisms [3] after they have entered the aquatic environment.

The selective serotonin reuptake inhibitors (SSRIs) are members of the most prescribed class of antidepressants in the USA and Europe [4–6]. They have been repeatedly detected in WW, surface waters, sediments, and aquatic organisms [7–12], and are thus part of different monitoring programs [13]. In aquatic organisms, SSRIs cause changes in biochemical processes, feeding behavior, survivorship behavior, growth, and potential changes in

**Citation:** Gornik, T.; Shinde, S.; Lamovsek, L.; Koblar, M.; Heath, E.; Sellergren, B.; Kosjek, T. Molecularly Imprinted Polymers for the Removal of Antide-Pressants from Contaminated Wastewater. *Polymers* **2020**, *13*, 120. https://doi.org/ 10.3390/polym13010120

Academic Editor: Marta Otero Received: 24 November 2020 Accepted: 23 December 2020 Published: 30 December 2020

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

**Copyright:** © 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 (https:// creativecommons.org/licenses/by/ 4.0/).

their genetic material [7,14–17]. Hence, it is crucial to improve their removal from WW before they are introduced into the environment. Among the existing WW treatment techniques, advanced oxidation processes and biological treatment are most successful in removing SSRIs from WW [18,19]. While during the former, the leading process of removal is degradation, sorption to activated sludge seems to be responsible for the removal of the majority of SSRIs during biological treatment [19]. Hence, other adsorption-based treatment techniques have been considered. Among them, activated carbon (AC) is by far the most researched material for SSRI removal [20,21]. AC as a treatment technique is technologically simple, has relatively fast kinetics, and removes a high variety of contaminants. Its main disadvantages are a high initial investment, the non-selectivity of the process, and the need for frequent regeneration due to fouling, which is expensive, time-consuming, and results in the loss of material in each regeneration cycle [22,23]. Greener alternatives, such as using products of pyrolysis of primary and secondary paper mill sludge, spent coffee grounds, and pine bark have been reported [24,25]. However, there is a lack of literature investigating modified synthetic composite materials, such as carbon-based nanomaterials, different types of membranes, and other forms of modified polymers, which, however, present promising alternatives to achieve superior SSRI removal from WWs [19,26–29]. On the basis of this knowledge gap, we investigated molecularly imprinted polymers (MIPs) as an alternative sorption material to AC [23,30,31].

MIPs are polymers that have been imprinted by a chosen template during the polymerization step in order to create selective recognition sites and are therefore often referred to as artificial antibodies or synthetic receptors [32]. After the template is removed from the MIP, the same or similar molecule can be rebound. They have already been commercially used for solid-phase extraction (SPE) [33,34] and researched for several other applications, such as catalysis, chromatography, and drug delivery [35,36]. In the last few years, the number of studies considering MIPs for water treatment has increased. Thus far, they have been utilized to remove non-steroidal anti-inflammatory drugs, antibiotics, antimicrobials, endocrine-disrupting compounds, herbicides, phenols, and beta-blockers from contaminated WW [30,35,37–41]. The advantages of using MIPs for water treatment are their high selectivity and affinity for their targets. Hence, we expect to be able to regenerate the material after longer intervals compared to AC, since slower fouling rates are expected. Literature reports MIPs as mechanically and chemically stable, and thus they should withstand several regeneration cycles unchanged, making the treatment more cost-effective [35,38]. The main disadvantage of MIPs is, however, the initial investment into the production of the polymers. Among multiple polymerization procedures available today, we chose bulk polymerization as one of the simplest and cheapest one for MIP production [2].

The aim of this work was to develop a MIP that could be used for removal of not only our targeted template, but for the whole class of SSRIs. We evaluated the affinity, capacity, and selectivity of the synthetized MIPs for sertraline (SER) and chose the best performing materials. Further characterization included cross-reactivity towards other antidepressants fluoxetine (FLU), paroxetine (PXT), escitalopram (ESC), bupropion (BUP), two SER metabolites—norsertraline (NS) and sertraline ketone (SEK) [9,10,42], and structurally related compound bupivacaine (BUC) (Figure 1). Potential parameters influencing the removal were considered and the performance of the MIPs in WW was tested in order to evaluate their applicability for WW treatment. The composition of the polymers was confirmed using Fourier transform infrared spectroscopy (FTIR) and elemental analysis. Surface properties and pore volume were calculated on the basis of the obtained Brunauer– Emmett–Teller (BET) isotherms, and scanning electron microscopy images of materials were taken for morphological characterization.

**Figure 1.** Chemical structures of the tested compounds.

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

The list of chemicals, materials, and the description of standard solution preparation and pre-preparation of the polymerization ingredients are reported in the Supplementary Material (SM) Section 1.

#### *2.1. The Synthesis of MIP*

The polymers were prepared via bulk radical polymerization with the ingredients in ratios specified in Table 1.

**Table 1.** Polymer compositions (molar ratio) and ingredients used for the synthesis of molecularly imprinted polymers (MIPs).


The mini-MIP library was synthesized by varying functional monomer, porogen, and the form of the template, as illustrated in the Table 1. The molar ratio between the template, functional monomer, and cross-linker was 1/4/20. In the case of mini-MIPs, 34.1 mg (0.1 mmol) of sertraline in HCl salt form (SER HCl) or 30.8 mg (0.1 mmol) free base sertraline (SER), 34 µL (0.4 mmol) methacrylic acid (MAA), and 380 µL (2 mmol) ethylene glycol dimethacrylate (EGDMA) was used. A total of 560 µL of porogen (either CHCl3, methanol—MeOH,

or acetonitrile—ACN) was added, with the exception of anhydrous toluene, where 580 µL was needed due to solubility issues. For polymers prepared using two functional monomers, we changed the ratio to 1/4/8/12 for the template (30.8 mg SER, 0.1 mmol), functional monomer (34 µL MAA, 0.4 mmol), co-monomer (860 µL of methyl methacrylate (mMA) or 970 µL of 2-hydroxyethyl methacrylate (HEMA), 0.8 mmol), and 227 µL of the EGDMA cross-linker (1.2 mmol). We used 1 wt % of the initiator 2,2'-azobis(2,4-dimethyl valeronitrile) (V-65) for synthesis of polymers on the basis of total monomers.

The synthetic procedure was identical for all MIPs. The monomers and the template were first mixed and dissolved in the porogen solvent. Then cross-linker EGDMA was added and the solution was mixed again. Finally, the initiator V-65 was added. The solution was mixed, purged with N<sup>2</sup> for 10 min, and polymerized at 50 ◦C for 24 h in an oven. After 24 h, the polymerization was carried out for another 2 hours at 70 ◦C. The corresponding non-imprinted polymers (NIPs) were prepared following the identical procedures in the absence of the template.

Best-performing MIPs and their corresponding NIPs were later prepared in a 10 times larger quantity, maintaining the same polymer compositions and ingredients. The polymers were then crushed and sieved into 25–50 µm particle size. Both MIPs and NIPs underwent Soxhlet extraction in 10% of acetic acid in methanol for 96 h until no SER was detected by a high-performance liquid chromatograph coupled with a diode array detector (HPLC-DAD). The polymers were further washed with water and MeOH to remove the acetic acid, before drying them in the oven at 50 ◦C for 24 h. The dried polymers were used for further physical and analytical characterization.

#### *2.2. Selection of the Material: Batch Rebinding*

Batch rebinding tests were performed in both water and acetonitrile (ACN). A total of 5 mg of each MIP and the corresponding NIP was weighed and placed in 1.5 mL Eppendorf tubes containing 500 µL of the SER solution with increasing concentrations: 0.1, 0.4, 1.0, 2.0, 3.0, and 4.0 mM. We used SER × HCl for rebinding in water, and SER in the free base form for the rebinding in ACN. All the experiments were performed after the equilibrium had been reached, i.e., after 20 h (see Section 2.5). The suspension was centrifuged at 10,000 rpm for 15 min. The supernatant was diluted 10 times with the mixture of 50% ACN and 50% 20 mM phosphate buffer at pH 3.70 (mobile phase) and subsequently quantified by HPLC-DAD analysis. The levels of bound compounds to the MIP/NIP for each solvent mixture were estimated from plotted calibration curves. We plotted the data in the form of rebinding isotherms using the bi-Langmuir isotherm as the best fit (*R* <sup>2</sup> > 0.90). The capacity, affinity, and selectivity were calculated for each polymer. Capacity was reported as the mass of bound compound per gram of polymer. Affinity was determined as the distribution ratio (D), the ratio between the amount of SER bound to the polymer (B), and the remaining SER in the supernatant (F). The selectivity was calculated as the imprinting factor (IF), comparing the D of MIP to the D of its corresponding NIP. All the parameters were calculated at equilibrium at the highest added concentration of 4.0 mM. On the basis of the results in both ACN and water, we chose three best performing MIPs for further testing.

#### *2.3. Reusability Experiments*

Reusability of the chosen MIPs and NIPs was tested by repeating 4 times the batch rebinding of 0.1 mM SER in ultrapure water (UW) on the same material, while following any changes in the performance. Between the cycles, the polymers were washed with 1 mL 1% trifluoroacetic acid (TFA) in MeOH (30 min) and 1 mL of MeOH (15 min) in order to remove SER from polymers. Solvent-free polymers were obtained by drying in the oven for 1 h at 60 ◦C. The experiment was performed in 5 parallels.

#### *2.4. Cross-Reactivity Experiments*

The cross-reactivity of the 3 materials selected as described in Section 2.2 was evaluated by binding experiments for antidepressants and their structurally related compounds: NS,

SEK, FLU, ESC, PXT, BUP, and BUC. The cross-reactivity was assessed through selectivity factor (α), the capacity, and the difference in binding between MIP and NIP for each compound. A was calculated as the ratio between the D of SER and D of the tested compound.

The cross-reactivity experiments were performed separately for each compound in UW, applying the same conditions as for SER rebinding tests (see Section 2.2.). The experiments were performed at the concentration of 1 mM, which was selected on the basis of the maximal solubility of NS in UW. SEK binding was evaluated in ACN due to solubility limitation. The concentrations in the supernatant were again determined with the HPLC-DAD.

#### *2.5. Time to Reach Equilibrium*

The time to reach the equilibrium state was estimated in batch experiments in UW. A total of 5 mg of each chosen polymer and AC were shaken for 15 min, 30 min, 1 h, 4 h, 8 h and 20 h. The 0.5 mL solutions contained a mixture of SER and the compounds included in Section 2.4 (test mixture), each added at the final concentration of 0.1 mM. The removal percentage was determined by HPLC-DAD.

#### *2.6. Binding in WW Matrix: Influence of pH, Salts, and Chemical Oxygen Demand*

The behavior of the chosen polymers and AC was observed in WW matrix spiked with the test mixture, again at the final concentration of 0.1 mM. The binding experiments were performed in 3 different matrices: UW, artificial wastewater (WW1) [43], and actual wastewater (WW2) obtained from a Slovenian wastewater treatment plant (WWTP). The WW was filtered (see SM, Section 1.1) before spiking in order to remove particulates and microorganisms that could have influenced the removal. The pH of the WWs was measured using the pH electrode by Wissenschaftlich-Technische Werkstätten GmbH (Weilheim, Germany) and the chemical oxygen demand (COD) was determined on a spectrophotometer using Hach reagents for water analysis, LCK 314 and 514.

We researched the influence of 2 parameters most often reported to influence the binding: pH and the presence of salt ions [44–46]. Since the reported pH of WW is between 6 and 8, the performance of the polymers was tested by batch tests in 50 mM phosphate buffer solutions with pH adjusted to 6.0, 7.0, or 8.0 with either a 2 mM HCl or 1 mM NaOH solution. The influence of salt ions was observed by comparing the binding in UW and in NaCl solutions at the concentrations of 0.1 M and 1.0 M.

#### *2.7. Upscale Experiment*

In order to observe the performance of the materials on a larger scale and at lower concentration of substrate, we packed the material into SPE cartridges by separately weighing 50 mg of MIP, NIP, or AC. MIPs and NIPs were sedimented beforehand in a mixture of MeOH and water (*v/v* = 80/20) four-times for 1.5 h to avoid the loss of material through the frit. For the same reason, AC mesh size 100–400 was used.

The materials were first washed with 5 mL of MeOH and 5 mL of UW water. Then, the cartridges were stacked on top of Oasis HLB cartridges in order to bind the remainder of the unbound compounds. The method used for Oasis HLB conditioning, equilibration, loading, and elution was adapted from our article on photodegradation of SER [9].

A total of 50 mL of WW2 spiked with the mixture of compounds at concentrations of 0.4 µM was loaded at the flow rate of 2 mL min−<sup>1</sup> on to each material. The solution then flowed directly onto the Oasis HLB cartridge. After loading, the Oasis HLB cartridges were dried for 30 min and then eluted with 3 × 0.6 mL of triethylamine in MeOH. The elution solvent was evaporated, and the extracts were redissolved in 0.5 mL the HPLC mobile phase and filtered through 0.45 µm syringe filters before the HPLC measurements.

#### *2.8. Leaching Evaluation*

To examine the applicability of developed MIPs as SPE extraction materials, we checked the potential leaching of the template from the MIP. As reported under the upscale experiment (Section 2.7), 50 mg of each MIP was packed in the SPE column, conditioned, loaded,

and eluted with 5 mL 1% TFA in MeOH. The extract was dried under nitrogen at 40 ◦C and the amount of leaching was quantified with a Nexera X2 ultra high performance liquid chromatograph (UHPLC, Schimadzu, Kyoto, Japan) coupled to the hybrid quadrupolelinear ion trap mass spectrometry analyzer QTRAP 4500 (Sciex, Framingham, MA, USA) following the method developed by Gornik et al., (2020a) [9].

#### *2.9. Chemical and Morphological Characterization*

Fourier transform infrared (FTIR) spectroscopy was performed on IRAffinity-1S (Schimadzu, Kyoto, Japan).

Elemental analysis was performed on a 2400, Series II, CHNS/O Analyzer (Perkin-Elmer, Waltham, MA, USA).

BET surface area analysis was performed with Porozimeter TriStar II (Micromeritics, Norcross, GA, USA).

The morphological characteristics were observed using a scanning electron microscope (SEM). The images were recorded with JSM-7600F (JEOL Ltd., Tokyo, Japan).

#### *2.10. HPLC Measurements*

For the determination of SER, NS, SEK, FLU, ESC, PXT, BUP, and BUC in the solutions, we utilized an HPLC-DAD (1260 Infinity Agilent Technologies, Santa Clara, CA, USA). For separation, we applied the column Zorbax Eclipse C-18 column (150 mm × 4.6 mm, 5 µm) (Agilent Technologies, Santa Clara, CA, USA). The injection volume was 10 µL or 20 µL, depending on the tested concentration range. The mobile phases were (A) ACN and (B) 20 mM phosphate buffer at pH 3.70. The gradient started with 70% B for 2 min, decreased to 61% in 13 min, then increased back to 70% B in 0.1 min and was kept as so for 1.5 min. The flow rate was 1 mL·min−<sup>1</sup> . The retention times of the compounds were 3.27 min for BUP, 4.04 min for BUC, 6.20 min for ESC, 8.52 min for PXT, 11.95 min for NS, 12.65 min for FLU, and 13.04 min for SER. SEK was determined with a separate method at flow 2 mL·min−<sup>1</sup> , isocratic elution at 70% A and 30% B. Other parameters coincided with the previous method. SEK eluted at 3.80 min.

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

All experiments with the exception of the reusability experiments (*n* = 5) were performed in duplicate. The inter-day repeatability reported as the relative standard deviation (RSD) for experiments performed in UW was <5% and in WW < 6%.

#### *3.1. MIP Synthesis, Selection, and Reusability*

We optimized the polymer composition to tune recognition properties of the material. The initiator (V-65) and cross-linker (EGDMA) were kept constant for all polymerization experiments, while different porogens and co-monomers were added in order to obtain water compatibility, increase capacity, and improve selectivity. The behavior of MIPs compared to their corresponding NIPs was evaluated in batch rebinding experiments performed in water and ACN at different concentrations to generate binding isotherms and calculate the capacity, affinity, and IF. The data we obtained during ACN rebinding experiments enabled us to quantify the binding on the basis only of specific interactions, such as hydrogen bonding, with the minimal non-specific hydrophobic effect [47], while our prime goal was recognition of the investigated compounds in water.

EGDMA in combination with MAA in different porogen solvents is one of the most commonly reported compositions of MIPs to date [30,48,49], including those in MIPs imprinted with SER [50,51]. Unlike in the literature [50,51], we observed no imprinting in MIPs where SER was used in its salt form (MIPs 1–3 in Table 1). Adding the extracted free base form of SER, on the other hand, resulted in successful imprinting. As shown in Figures 2 and 3, we observed higher capacities and affinities in water compared to those in ACN for most tested MIPs, except for MIP11 and MIP12. However, compared to ACN, the Ifs in water were lower in all the cases, indicating loss of selectivity in water. This can be justified

by the hydrophobic effect established in polar solvents such as water, and disrupting the formation of hydrogen bonds. In order to improve the recognition abilities in water, we tested the influence of adding co-monomers mMA or HEMA (see examples MIP8– MIP13 in Table 1). Here, the ratios between monomers and cross-linker we applied were based on the results from Dirion et al., (2003). HEMA was chosen on the basis of the reports on improved Ifs in water [47,48], while the mMA was selected as its more non-polar alternative. MIPs with the mMA added into the polymerization mixture (MIP8, MIP9, and MIP10) had a similar IF in ACN, as compared to MIPs 5–7, which were prepared by MAA only (Table 1). However, the Ifs in water were slightly higher for all three materials (Figure 3). The capacities and affinities of MIP8, MIP9, and MIP10 in ACN were higher, yet lower or comparable in water. Compared to MIPs 5–7, adding HEMA as a co-monomer (MIP11, MIP12, and MIP13) did not improve the capacity or affinity of the MIPs in ACN. Additionally, both parameters were noticeably lower in UW. The considerable improvement was, however, observed in IF; the highest was that of MIP13. This high IF is in agreement with the results of Dirion et al., (2013).

**Figure 2.** Rebinding isotherms for MIP (blue symbols) and non-imprinted polymer (NIP; red symbols) combinations 5–13 in ultrapure water (UW; full symbols) and acetonitrile (ACN; empty symbols).

− − − − **Figure 3.** (**a**) The selectivity (imprinting factor, IF), (**b**) the affinity (L·g −1 ), and (**c**) the capacity (mg·g −1 ) of the MIPs in UW in blue. (**d**) The selectivity (IF), (**e**) the affinity (L·g −1 ), and (**f**) the capacity (mg·g −1 ) of the MIPs in ACN in green.

As seen in Table 1 and Figure 3, the porogen severely influenced the selectivity, capacity, and affinity of the MIPs. This happens as it affects the stability of the "prepolymerization complex" (i.e., interactions between functional monomers and the template in the chosen porogen), which plays a crucial role in the imprinting effect. If the porogen disrupts hydrogen bonds between the template and monomers, no specific binding is observed, as can be seen in the case of MeOH (MIP4). On the contrary, using a more nonpolar aprotic porogen, the pre-polymerization complex is stabilized, resulting in higher IF, which we showed in MIPs 7, 10, and 13 synthetized in toluene, as compared to those synthetized in ACN (MIPs 6, 9, 12) or CHCl<sup>3</sup> (MIPs 5, 8, 11) (Table 1, Figure 3a,d) [47].

Determining rebinding characteristics allowed us to select three most promising materials for further testing. In terms of capacity and affinity in water, the material MIP5 was chosen. MIP13 was chosen for its highest IF. Lastly, MIP9 was chosen because it combines satisfactory selectivity, capacity, and affinity in both water and ACN. The chosen polymers were reusable, with the maximum observed decrease in the capacity for SER in four consecutive rebinding experiments being only 2%.

#### *3.2. Cross-Reactivity*

We determined the cross-reactivity of three selected MIPs for the following antidepressants and structurally related compounds (Figure 1): BUC, BUP, ESC, PXT, NS, SEK, and FLU. While SEK, the metabolite of SER, was also initially included, it however showed very poor binding in ACN and no observed selectivity for any of the three MIPs. Its binding will therefore be based on non-specific interactions only. As for its poor solubility in water, it was thus excluded from further testing.

α α α The results of cross-reactivity tests are selectivity factors (α) reported in Table 2. In general, α for each compound were comparable between the three selected MIPs, with the exception of BUP in MIP13. Here, the factor α 3.29, as compared to 9.60 and 9.76 for MIP5 and MIP9, respectively, indicated more cross-reactivity of MIP13 towards BUP. As reported in Table 2, for NS the selectivity factor was below 1, indicating better

binding as compared to SER, which is reasoned by the absence of the methyl group in the chemical structure (Figure 1). Among the SSRI compounds, FLU and PXT had the factors slightly above 1, meaning comparable binding, while the factor for ESC varied between 2.7 and 2.9. The fact that ESC was the only SSRI with a tertiary amine in the structure, together with the favorable α for NS, suggests the impact of steric hindrance of the hydrogen bond-forming amino group on cross-reactivity. The size of the binding site seemed to be of lesser importance, considering that PXT and FLU are larger molecules as compared to SER and NS. BUC and BUP showed higher α in all three MIPs, which is justified by them being less structurally related to the SSRI group. Additionally, their amino groups are also sterically more hindered (tert-butyl group and tertiary amine).

**Table 2.** The capacity and selectivity factor of MIP5, MIP9, and MIP13 and the difference in binding of each compound between MIP and the corresponding NIP at 1 mM concentration of each tested analyte.


In general, the capacities of the three MIPs for SSRIs followed the same pattern as in SER binding. The highest capacity was observed in MIP5, closely followed by MIP9, and with more than half-lower capacities observed in MIP13 (Figure S1). Furthermore, we compared the binding to the corresponding NIPs. The difference between MIP and NIP was the largest in the case of MIP/NIP13 and the lowest in MIP/NIP5 (Table 2).

#### *3.3. Time to Reach the Equilibrium*

The time to reach equilibrium was tested for the chosen polymers and AC. A 0.1 mM test mixture was added. Figure S2 illustrates that for AC, the equilibrium was reached within 1 h; in cases of MIP5 and MIP9, the equilibrium was reached in 4 h; and for MIP13, in 20 h. For the NIPs, similar times to reach the equilibrium were shown as for their corresponding MIPs. As also depicted from Figure 4, AC non-selectively bound all the available compounds until their concentrations in the solvent reached below the limit of quantification (LOQ ≈ 0.001 mM).

**Figure 4.** The impact of matrices (UW, artificial wastewater (WW1), and actual wastewater (WW2)) on the performance of MIPs, NIPs, and AC in the batch rebinding test.

#### *3.4. Effect of WW Matrix*

− − − − ′ With the underlying objective to remove pharmaceuticals from WWs, we tested the capacity of MIPs to bind them. This way, we evaluated the ability of MIPs to be applied as sorbents in WW treatment systems. Aiming to get closer to the conditions during WW treatment, we employed the pH adjusted to 6–8 and simulated actual WW matrix composition. By comparing the results between the binding of BUP, BUC, ESC, PXT, NS, FLU and SER in UW, WW1, and WW2, we observed large differences in the removals of the test compounds (Figure 4), whereas AC removed all the tested compounds in any matrix to below LOQ concentrations (0.001 mM). In contrast with our expectations, as shown in Figure 4, the removal efficiencies of MIPs were lowest in UW and highest in the most complex matrix, WW2. In line with the trends shown in the capacity experiments, MIP5 and MIP9 showed best performance, closely followed by NIP5, NIP9, MIP13, and finally NIP13. By investigating the reason for such behavior, we determined the pH and COD of each inspected matrix. The pH values of UW, WW1, and WW2 were approximately 7, 7.2, and 8.2, respectively, whereas we measured COD at <15 mg·L −1 for WW1 (LOQ of the test) and 379 mg·L −1 for WW2. On the contrary, the literature reports either no change (up to 690 mg·L <sup>−</sup><sup>1</sup> COD) or a slight decrease in adsorption of their chosen templates to their MIPs at high COD values (over 800 mg·L −1 ) [52–55]. The MIPs in these cases used similar reagents to those in our synthesis, i.e., MAA and EGDMA, albeit in different ratios, and employed DCM or ACN as porogens and 2,2′ -azobisisobutyronitrile (AIBN) as the initiator [52–55]. Hence, we did not expect the higher COD values to be the cause behind the increased removal.

In order to deeper investigate the reasons behind the positive impact of matrix complexity on the removal of pharmaceuticals, we performed the rebinding experiments at different pH values and salt concentrations. Here, the imprinted and non-imprinted polymers showed similar trends, with the most notable differences for MIP and NIP13, as portrayed in Figure 5. The pH in the range of 6 to 8 had almost no influence on the bind-

ing with differences below 1%. The only exception was NIP13, with differences between pH 6 and pH 8 ranging up to 7.8%. On the other hand, the increasing salt concentration improved the removal of pharmaceuticals. This finding was further supported by the improved binding found during the pH tests, which were performed in phosphate buffer, as compared to the binding in UW. Our results are consistent with the findings of Kempe and Kempe (2010), where elevated concentrations of salts had a significant influence on the removal of penicillin G from solution and followed the Hofmeister series. As seen in Kempe and Kempe (2010), the higher removal was of non-specific nature, observed in both MIP and NIP [46]. The kosmotropic ions seem to promote the formation of stable interactions between the polymers and tested compounds. Since phosphate ions are more kosmotropic than chloride ions, this would also explain the larger effect in the buffer solutions, despite their lower concentrations [56].

**Figure 5.** The effect of pH and presence of salt ions in MIP13 and NIP13.

#### *3.5. Upscale Experiment*

The performance of the materials was evaluated as the difference between the initial concentration and the remainder extracted by Oasis HLB SPE. This way, we avoided underestimating the performance of AC, since completely eluting compounds off the AC is a known difficulty [23]. The results on the performance of selected materials in the upscale experiment are shown in Figure 6. The main difference from the batch (mini-MIP) experiments is the less efficient binding to AC (Figure 6). The two main reasons behind this may involve the shorter contact time between the material and WW, or lower capacity of the material due to the non-specific binding of other matrix components. Since in the batch experiment AC showed shortest time to reach equilibrium, the latter is more probable. Furthermore, several reports showed AC performance deteriorating with an increase of matrix complexity (e.g., COD, total dissolved solids) [23,55].

**Figure 6.** The remainder of compounds detected in the Oasis HLB extracts in the upscale experiment.

In the Oasis HLB extracts from MIP5, NIP5, and MIP9, none of the investigated pharmaceuticals were detected. On the contrary, as expected, their highest remainder was determined in the Oasis HLB extracts from NIP13, again implying its lowest binding capacity.

While non-specific rebinding is not desired in MIPs that are used, for example, in sample preparation or chromatography, we show here that this phenomenon is favorable in WW treatment. As Le Noir et al., (2007) pointed out, it only becomes a problem if it causes lower capacity and affinity of the selective binding [39]. MIP5 and MIP9 both showed higher capacities compared to MIP13, and even NIP5 and NIP9 performed better under tested conditions. This means that a larger amount of MIP13 would have to be used to achieve the competitive removal efficiencies. However, specific interactions of MIPs will likely play a more important role at higher volumes and more complex matrices. At the same time, we show that the NIPs, which are based on non-specific binding only, are less negatively affected by matrix, as compared to AC, and along with their easy recyclability they could therefore pose a less expensive alternative for the removal of pharmaceuticals.

#### *3.6. Leaching*

As an alternative to sorption in WW treatment, we also considered the developed MIPs for SPE extraction of environmental samples. As for our hypothesis, MIP could be employed as an SPE sorbent in order to selectively extract targeted compounds, thus reducing the suppressing effect of matrix interferences in further liquid chromatography coupled to mass spectrometry (LC–MS) analysis. Such sorbents may potentially be employed in a highly sensitive analytical method for an ultra-trace level determination of contaminants in WW [57]. MIPs have previously been used for SPE several times [58–61]. However, given the fact that the template in polymerization (SER) is also the analyte in the LC–MS method, the MIP sorbent would have to pass the "leaching test", which means that it would have to show a negligible leaching and thus avoid interfering with the assessment of trace-level analytes in the subsequent LC–MS analysis. Leaching of SER from the material was tested on UHPLC-QTRAP, applying the instrumental method developed by Gornik et al., (2020a) [9]. By using 5 mL of 1% TFA in MeOH, we eluted up to 3.5 µg of SER from the MIPs. Alternative methods for template removal, such as microwave or ultrasound-assisted extraction, heating under pressure, or even the use of another acid during Soxhlet extraction, could have lessened the leaching from the MIPs. On the other hand, the more extreme conditions could also have damaged or distorted the imprinted cavities and thus decreased the selectivity, affinity, and capacity of the MIPs [62,63]. Furthermore, the synthesis of MIPs and the subsequent washing procedures triggered the formation of SER transformation products (NS, SEK, hydroxyl-SER) [9], which in turn leached off the materials, thus interfering the environmental analysis. Unfortunately, this makes the material inappropriate for the determination of SER residues including its metabolites and transformation products at trace levels. Finding an appropriate dummy template that would substitute SER and produce a MIP cross-reactive towards SSRI could be a viable

solution to such a problem [35]. Nonetheless, the synthetized material can still be applied to SPE of the remaining tested pharmaceuticals (Figure 1).

#### *3.7. Characterization*

The FTIR spectra for the chosen MIP/NIP pairs 5, 9, and 13 can be found in Figure 7. The broad band visible at approximately 3500 cm−<sup>1</sup> corresponds with the stretching vibration of the hydroxyl group from MAAs COOH group. The stretch bands around 2950 cm−<sup>1</sup> in all the spectra are part of the C–H vibration present in MAA, mMA, HEMA, and EGDMA. The band around 1720 cm−<sup>1</sup> represents the vibration from the carboxylic C=O group that can be associated with the C=O groups from MAA, mMA, and EGDMA. The 1250 and 1140 cm−<sup>1</sup> stretch bands contributed to the stretching of C–O also present in all three compounds. The stretch bands corresponded with the polymerized material. Since the composition of the synthetized materials did not vary strongly, the resulting FTIR spectra were accordingly similar.

**Figure 7.** The FTIR spectra of MIPs and NIPs 5, 9, and 13.

The results of the elemental analysis of the MIP and NIP pairs 5, 9, and 13 are reported in Table 3. The results are in accordance with the expected values of the synthetized material. With this measurement, we confirmed that the added reagents reacted in the expected ratio.


**Table 3.** Results of the elemental analysis for MIPs and NIPs 5, 9, and 13.

The BET surface area, pore size, and pore volume of the MIPs and NIPs are reported in Table 4. As expected, the larger the surface area and pore volume of the tested polymers, the higher the reported capacity and affinity. All three parameters were comparable between MIP and NIP pairs 5 and 9, with BET surface areas for MIP/NIP 5 in the 200 m<sup>2</sup> ·g −1 range and MIP/NIP 9 at the 100 m<sup>2</sup> ·g −1 range. However, NIP13 exhibited a more than five times lower BET surface area and pore volume compared to its corresponding MIP (Table 4). A similar difference was observed in MIP and NIP pairs using HEMA as the copolymer in toluene in the research by Dirion et al., (2003). They reported that stronger swelling was observed for the NIPs and similar elution times measured for void markers (acetone or MeOH) in their chromatographic evaluations of the polymers. This indicated a smaller difference between the MIP and NIP in their swollen state.


**Table 4.** Brunauer–Emmett–Teller (BET) surface area of MIPs and NIPs 5, 9, and 13.

The SEM images of the surface of our polymers in Figure 8 support the surface area and pore volume measurements. While the morphology of MIP5/NIP5 and MIP9/NIP9 were comparable, the surfaces of MIP13 and NIP13 were dissimilar. These differences in the morphology between MIP and NIP 13 indicate that care should be taken when NIPs are used for the evaluation of MIP selectivity. Comparing a material imprinted with a completely different compound or the determination of α between the template and other compounds can offer more information [35,48].

**Figure 8.** SEM images of MIPs and NIPs 5, 9, and 13.

− Compared to AC with a surface area of 1400 m<sup>2</sup> ·g −1 [64], the surface areas of MIPs and NIPs were 5 to 253 times lower. Nevertheless, some of them showed superior binding characteristics in WW.

#### **4. Conclusions**

This study investigated the ability of MIPs imprinted with the free base form of SER to remove SSRIs and their metabolites. The functional monomers and porogens revealed a strong impact on the capacity, affinity, and selectivity of the synthetized MIPs. The three selected MIPs showed cross-reactivity towards the SSRIs and the metabolite norsertraline, whereas they bound a lesser amount of the competitors BUP and BUC. Further, the loss of selectivity towards the metabolite SEK was probably due to the loss of the amino group, which was thus

found crucial for selective binding to the MIP. The performance of both the imprinted and non-imprinted materials was strongly influenced by the presence of salt ions, which improved their performance in WW. The performance of MIPs was stable throughout WW-relevant pH range 6–8. Compared to AC, the synthetized polymers had at least five times lower surface area and required a longer equilibration time. This slower mass transfer was particularly evident when selective binding was the main driving force behind the removal, as observed in MIP13. However, the capacity in WW for two out of the three tested MIPs surpassed that of AC, and thus both the non-specific and specific interactions showed an important role for the removal from WW. The surface area calculated from the BET isotherm for the MIPs correlated with a higher removal and more non-specific interactions. The advantage of the MIPs is also their reusability that, together with the lower number of regeneration cycles needed due to slower fouling, will cut the costs of the treatment. Unfortunately, the MIPs were found inappropriate for SPE of samples containing trace levels of SER due to continuous leaching of the template and its degradation products. Future work should include a large-scale experiment confirming the advantages of the synthetized material for the removal of SSRIs from WW.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2073-4 360/13/1/120/s1: List of standards, chemicals, and materials; Standard solution preparation; Prepreparation of the ingredients. Figure S1: Cross-reactivity of MIP5, MIP9, and MIP13 in UW. Figure S2: Time to reach the equilibrium for MIP5, MIP9, MIP13, and activated carbon (AC).

**Author Contributions:** Conceptualization, T.G., S.S., B.S. and T.K.; methodology, T.G., S.S., B.S. and T.K.; analysis and investigation, T.G., L.L., M.K. and S.S.; writing—original draft preparation, T.G.; writing—review and editing, S.S., B.S., T.K. and E.H.; supervision, B.S. and T.K.; project administration, B.S. and T.K.; funding acquisition, B.S., E.H. and T.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** The authors acknowledge the financial support from the Slovenian Research Agency (research core funding no. P1-0143) and project J1-6744 (Development of Molecularly Imprinted Polymers and their application in environmental and bio-analysis). This work was supported by a STSM Grant from the NEREUS COST Action ES1403 and the Erasmus plus program.

**Acknowledgments:** Special thanks go to Amadeja Koler from the University of Maribor, Faculty of Chemistry and Chemical Engineering, PolyOrgLab, Maribor, Slovenia, for the work on chemical characterization of the polymers.

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

#### **References**


*Article*

### **Chain Entanglement of 2-Ethylhexyl Hydrogen-2-Ethylhexylphosphonate into Methacrylate-Grafted Nonwoven Fabrics for Applications in Separation and Recovery of Dy (III) and Nd (III) from Aqueous Solution**

### **Hiroyuki Hoshina \*, Jinhua Chen \* , Haruyo Amada and Noriaki Seko**

Department of Advanced Functional Materials Research, Takasaki Advanced Radiation Research Institute, Quantum Beam Science Research Directorate, National Institutes for Quantum and Radiological Science and Technology, 1233 Watanuki-machi, Takasaki, Gunma 370-1292, Japan; amada.haruyo@qst.go.jp (H.A.); seko.noriaki@qst.go.jp (N.S.)

**\*** Correspondence: hoshina.hiroyuki@qst.go.jp (H.H.); chen.jinhua@qst.go.jp (J.C.); Tel.: +81-27-346-9125 (J.C.)

Received: 27 October 2020; Accepted: 9 November 2020; Published: 11 November 2020

**Abstract:** A nonwoven fabric adsorbent loaded with 2-ethylhexyl hydrogen-2-ethylhexylphosphonate (EHEP) was developed for the separation and recovery of dysprosium (Dy) and neodymium (Nd) from an aqueous solution. The adsorbent was prepared by the radiation-induced graft polymerization of a methacrylate monomer with a long alkyl chain onto a nonwoven fabric and the subsequent loading of EHEP by hydrophobic interaction and chain entanglement between the alkyl chains. The adsorbent was evaluated by batch and column tests with a Dy (III) and Nd (III) aqueous solution. In the batch tests, the adsorbent showed high Dy (III) adsorptivity close to 25.0 mg/g but low Nd (III) adsorptivity below 1.0 mg/g, indicating that the adsorbent had high selective adsorption. In particular, the octadecyl methacrylate (OMA)-adsorbent showed adsorption stability in repeated tests. In the column tests, the OMA-adsorbent was also stable and showed high Dy (III) adsorptivity and high selectivity in repeated adsorption–elution circle tests. This result suggested that the OMA-adsorbent may be a promising adsorbent for the separation and recovery of Dy (III) and Nd (III) ions.

**Keywords:** selective adsorption; dysprosium; neodymium; fabric adsorbent; radiation; graft polymerization

#### **1. Introduction**

Rare earths including scandium, yttrium, and 15 lanthanoid elements, have recently become indispensable materials for the high-tech industry. Due to the uneven distribution of rare-earth sources in the world, almost all rare earths are supplied by limited countries [1]. Therefore, it is necessary to recycle used rare earths to ensure a stable supply of these materials in many countries [2–7]. Among the rare earths, dysprosium (Dy) and neodymium (Nd) are listed as "critical materials" by the United States due to supply issues and their importance to electronics and electrical technology [8,9]. For example, neodymium and dysprosium are key components of permanent magnets, such as NdFeB magnets. The demand for the separation and recovery of dysprosium and neodymium from used permanent magnets and scraps generated during manufacturing is increasing [10–20].

The technology for the separation and recovery of dysprosium and neodymium from used permanent magnets has been extensively studied [21,22]. The most common method of recovering dysprosium and neodymium from waste materials involves leaching them in an acid solution and purifying the leached ions by solvent extraction [11–13,23]. Organophosphorus compounds such as 2-ethylhexyl hydrogen-2-ethylhexylphosphonate and di(2-ethylhexyl)phosphoric acid, carboxylic acid such as neodecanoic acid and naphthenic acid, and methyltrioctylamine chloride are usually used as extractants for rare-earth ions due to their good separation and recovery performance [22–26]. However, solvent extraction requires a large number of separation steps, a long processing time, and a large space for all necessary equipment. On the other hand, other methods such as chemical precipitation and ionic liquids extraction are also used for the separation and recovery of rare earths. Although the chemical precipitation process is simple and low in cost, the purity and recovery ratio of the resulting product are usually low, while the ionic liquid extraction cost is high for actual application [21]. Currently, the effective separation and recovery of rare earths from an aqueous solution requires relatively simple processes [27,28]. Adsorption techniques using adsorbents, such as inorganic particles, ion-exchange resins, and polymer ligands, are attractive for the separation and recovery of rare-earth ions [29–36]. This is because the adsorption process does not require much energy and water and can be easily operated anywhere by batch or column methods [37].

Inorganic particles, such as clay minerals, activated carbon, and magnetite nanoparticles, are highly suitable for removing heavy metals from water and wastewater. In many cases, these inorganic materials show high adsorption but low selectivity [37–39]. On the other hand, adsorbents with special ligands or chelating functional groups can be designed to selectively separate and recover target metal ions in water. These adsorbents, including ion-exchange resins and polymer ligands, can be prepared by introducing functional groups onto polymer materials by the radiation-induced graft polymerization method. This method can introduce new functional properties while maintaining the properties of the trunk polymers [40–48]. Various vinyl monomers have been radiation-grafted onto trunk polymers, such as polyethylene [41,42], polypropylene [43,44], fluoropolymers [45], and cellulose [46,47]. Furthermore, graft polymerization can be applied to various types of materials, such as films [45], fabrics [30,45–47], fibers [46], and particles [48]. Various adsorbents have been developed using this technology for the recovery and removal of metal ions from environmental water and industrial wastewater [46–51]. In the design of these adsorbents, it is important to select the most suitable functional groups based on the metal ion that needs to be adsorbed.

We noticed that 2-ethylhexyl hydrogen-2-ethylhexylphosphonate (EHEP), used as an extractant in the solvent extraction process, has two alkyl chains on each molecule [10,24,52]. In this study, we attempted to load EHEP onto polyethylene-coated polypropylene (PE/PP) nonwoven fabrics to develop a novel adsorbent for rare-earth ions. For this purpose, we grafted a polymerized methacrylate monomer with a long alkyl chain onto the fabrics. The EHEP was then loaded onto the grafted fabrics by hydrophobic interaction and chain entanglement between the alkyl chains. Here, since the EHEP is only physically bonded on the fabrics by hydrophobic interaction and chain entanglement, the loss of EHEP is a concern in practical applications. Therefore, the stability of EHEP-loaded adsorbents needs to be confirmed for practical use.

Four methacrylate monomers with different alkyl chain lengths—butyl methacrylate (BMA), hexyl methacrylate (HMA), dodecyl methacrylate (DMA), octadecyl methacrylate (OMA)—were radiation-grafted onto the PE/PP nonwoven fabrics in this study. The grafted fabrics were then loaded with EHEP to prepare the adsorbents. The adsorbents were tested in batch and column modes using Dy (III) and Nd (III) ion solutions [18]. The effects of the alkyl chain length of the monomers on the stability and adsorption performance of the EHEP-loaded absorbents were studied and evaluated.

#### **2. Experimental**

#### *2.1. Materials*

The trunk material used for graft polymerization was a nonwoven fabric composed of polyethylene-coated polypropylene (PE/PP) fibers, provided by Kurashiki Textile Manufacturing Co., Ltd., Kurashiki, Japan. The PE on the fiber surface is easy to be radiation-grafted, and the PP core makes the fiber mechanically stronger. Furthermore, the PE/PP nonwoven fabric is relatively cheap among artificial fabrics and has a large specific surface. The four methacrylate monomers—butyl methacrylate (BMA), hexyl methacrylate (HMA), dodecyl methacrylate (DMA), and octadecyl methacrylate (OMA)—are of chemical reagent grade and were purchased from Fujifilm Wako Pure Chemical Corporation, Tokyo, Japan. 2-Ethylhexyl hydrogen-2-ethylhexylphosphonate (EHEP) was provided by Daihachi Chemical Industry Co., Ltd., Tokyo, Japan. The other reagents, such as Tween 20 surfactant, methanol, ammonia water, HCl solution, Dy (III) (Dy2O<sup>3</sup> in 5 wt.% HNO3) standard solution, and Nd (III) (Nd2O<sup>3</sup> in 5 wt.% HNO3) solution, were purchased from Kanto Chemical Co., Inc., Tokyo, Japan. All chemicals were used without further purification. In this study, the deionized Mili-Q water with a high resistivity of 18 MΩ cm was used. Ω

#### *2.2. Graft Polymerization of Methacrylate Monomers*

Figure 1 shows the process of preparing the fabric adsorbents. Graft polymerization was performed using a preirradiation method. In this study, either PE nonwoven fabric or PP nonwoven fabric could be used as trunk polymers. However, the mechanical strength of common PE nonwoven fabric is significantly lower than that of PP nonwoven fabric, while the PP nonwoven fabric deteriorates faster than PE nonwoven fabric. Therefore, we chose the PE-coated PP nonwoven fabric as the polymer trunk for radiation grafting. The PE/PP nonwoven fabric with a size of 5 cm × 8 cm was placed in a polyethylene bag, purged with nitrogen gas to create an oxygen-free environment, and electron beam preirradiated at −80 ◦C (dry ice) with a beam energy of 2 MeV at a current of 3 mA to generate radicals on the fabric. The preirradiated fabric was removed and filled into a glass ampoule, which was evacuated and filled with a nitrogen-bubbled monomer solution to immerse the fabric completely. The ampoule was placed in a temperature-controlled oven. Under these conditions, the radicals initiated graft polymerization. The monomer structures and grafting conditions are shown in Table 1. After graft polymerization, the fabric was washed with methanol to remove residual monomers and homopolymers and dried in an oven at 60 ◦C for more than 24 h.

**Figure 1.** Preparation of the fabric adsorbents by graft polymerization of methacrylate monomers and the subsequent 2-ethylhexyl hydrogen-2-ethylhexylphosphonate (EHEP) loading.

The degree of grafting and the density of alkyl chains of the grafted fabrics were calculated using the following equations.

$$\text{Degree of grafiting (\%)} = (\text{W}\_{\text{g}} - \text{W}\_{\text{0}}) / \text{W}\_{\text{0}} \times 100 \tag{1}$$

$$\text{Density of alkali chains (mmol/g)} = 1000 \times (\text{W}\_{\text{g}} - \text{W}\_{0}) \text{M/W}\_{\text{g}} \tag{2}$$

where W<sup>0</sup> and W<sup>g</sup> are the dry weights (mg) of the fabrics before and after graft polymerization, and M is the molecular weight of the monomers as shown in Table 1.


**Table 1.** Methacrylate monomers and grafting conditions used in this study.

\* M is the molecular weight of the monomers; \*\* For the monomer solutions, monomer concentrations were fixed at 5.0 wt.% in water for BMA, HMA, and DMA, and in a water/methanol mixture solvent (1:1 in weight) for OMA; 0.5 wt.% of Tween 20 surfactant was added to the monomer solutions. \*\*\* Preirradiation was performed at −80 ◦C (dry ice) in an oxygen-free environment.

#### *2.3. Loading of EHEP onto the Grafted Fabrics*

A 50 wt.% EHEP solution of ethanol was uniformly dropped onto the grafted fabric for EHEP loading. The EHEP-loaded fabric was dried in a vacuum oven at 40 ◦C to remove the ethanol solvent. EHEP loading of the resulting fabric adsorbent was calculated by the following equation.

$$\text{HEEP loading (mmol/g)} = 1000 \times (\text{W}\_{\text{a}} - \text{W}\_{\text{g}}) \text{\textdegree 306/W}\_{\text{a}} \tag{3}$$

where W<sup>a</sup> is the dry weights (mg) of EHEP-loaded fabric, and 306 is the molecular weight of EHEP. The prepared fabric adsorbents with different monomers were named BMA-, HMA-, DMA-, and OMA-adsorbent, respectively.

#### *2.4. Characterization*

Fourier transforminfrared (FTIR) spectroscopic analysiswas performedwith an FTIR spectrophotometer in the attenuated total reflectance (ATR) mode (Spectrum One, PerkinElmer, Inc., Tokyo, Japan). The scanning range and resolution were 500–2500 cm−<sup>1</sup> and 1 cm−<sup>1</sup> , respectively.

The hydrophobicity of the grafted fabric was examined by measuring the contact angle with a contact angle meter (CA-X, Kyowa Interface Science Co., Ltd., Tokyo, Japan).

#### *2.5. Batch Adsorption Tests*

The prepared fabric adsorbent was evaluated by batch adsorption tests. The test solution contained 100 ppm Dy (III) and 100 ppm Nd (III). The pH of the test solution was adjusted to 2.0 by ammonia water. The fabric adsorbent with a size of 2 cm × 2 cm was immersed in 50 mL of test solution in a glass bottle. The bottle was placed on a shaker and shaken at a rate of 150 rpm at 25 ◦C for 3.0 h. After the adsorption test, the adsorbent was washed with deionized water to remove the unadsorbed ions on them.

To elute the adsorbed ions, the fabric adsorbent was immersed in 50 mL of 1.0 M HCl solution in a glass bottle, and the bottle was shaken at a rate of 150 rpm at 25 ◦C for 1.0 h. After elution, the fabric adsorbent was washed with deionized water and adsorption was repeated under the same conditions as the first adsorption test.

The ion concentrations in the adsorption and elution solutions were analyzed before and after each test with an inductively coupled plasma optical emission spectrometer (ICP-OES, Optima 8300, PerkinElmer, Inc., Tokyo, Japan). The adsorptivity (mg/g) of the fabric adsorbent was calculated as follows.

$$\text{AdSortivity } (\text{mg/g}) = 1000 \times (\text{C}\_0 - \text{C}\_i) \times \text{V/W}\_a \tag{4}$$

where C<sup>0</sup> (mg/mL) and C<sup>i</sup> (mg/mL) are the metal ion concentrations in the solution before and after the adsorption, respectively, and V (mL) is the volume of the solution.

#### *2.6. Column Adsorption Tests*

For the column adsorption tests, the fabric adsorbent with a diameter of 7.0 mm was packed into a column with an inner diameter of 7.0 mm. The volume of the adsorbent packed in the column was 0.2 mL. The test solution (100 ppm Dy (III) and 100 ppm Nd (III), pH 2) was passed through the column at a space velocity (SV) of 100 h−<sup>1</sup> at 25 ◦C. The SV is calculated by dividing the solution flow rate (mL/h) by the volume of adsorbent in the column (fixed at 0.2 mL in this study). A fraction collector was used to continuously collect the effluent from the column, and the ion concentrations were detected by ICP-OES. By plotting the relationship between C<sup>i</sup> and bed volume (BV), the ion concentration curve of the effluent was obtained. Here, C<sup>i</sup> is the ion concentration of the effluent at BV, and BV is calculated by dividing the total effluent volume from the column by the adsorbent volume (0.2 mL).

The adsorptivity (mg/g) of the adsorbent packedin the column was calculated by the following equation

$$\text{AdSortivity (mg/g)} = 1000 \times \sum \text{(C}\_0 - \text{C}\_i\text{)} \,\Delta \text{V}\_i / \text{W}\_a \tag{5}$$

where ∆V<sup>i</sup> (mL) and C<sup>i</sup> (mg/mL) are the volume and concentration of each collected effluent during the adsorption, respectively.

After the adsorption test, the adsorbent was thoroughly washed by passing deionized water through the column. Then, 1.0 M HCl solution of the eluent was passed through the column with a space velocity of 100 h−<sup>1</sup> at 25 ◦C until no metal ions were detected in the effluent. The eluted amount (mg/g) and recovery ratio were calculated by the following equations.

$$\text{Eluted amount (mg/g)} = 1000 \times \sum \text{C}\_{i} \,\Delta \text{V}\_{i}/\text{W}\_{\text{a}} \tag{6}$$

$$\text{Recovery ratio (\%)} = \text{Eluted amount} \vee \text{Adsorption} \times 100\tag{7}$$

where ∆V<sup>i</sup> (mL) and C<sup>i</sup> (mg/mL) are the volume and concentration of each collected effluent during the elution, respectively.

After the elution test, the adsorbent in the column was thoroughly washed with deionized water and used for the adsorption test again to evaluate its stability.

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

#### *3.1. Synthesis of EHEP-Loaded Adsorbent*

The adsorbent was prepared by the radiation-induced graft polymerization of methacrylate with a long alkyl chain onto PE/PP nonwoven fabric and the subsequent loading of EHEP by hydrophobic interaction and chain entanglement between the alkyl chains. Here, the EHEP organophosphorus compound has a special affinity for Dy (III) ions. The grafting results and the density of EHEP loading are summarized in Table 2.

**Table 2.** Degree of grafting and alkyl group density of the grafted fabrics, and the EHEP loading of the corresponding adsorbents.


\* Alkyl group density of the monomer-grafted fabric was calculated using Equation (2); \*\* EHEP loading was calculated by the weight increase of the grafted fabric before and after EHEP loading using Equation (3).

As shown in Table 2, four monomers with different alkyl chain lengths—BMA, HMA, DMA, and OMA—were radiation-grafted onto the fabrics. For comparison, the alkyl chain density in the grafted fabric was adjusted to be close to 2.0 mmol/g. For this reason, the degree of grafting was significantly different for each monomer and increased in proportion to the molecular weight of the grafted monomer. For example, to obtain a similar alkyl chain density of 2.0 mmol/g, the degree of grafting for the BMA is 51%, while it is 219% for the OMA. The latter is approximately four times higher than that of the former.

To obtain similar alkyl chain densities of the grafted fabrics, BMA grafting was carried out by immersing the 10 kGy preirradiated fabric into a 5.0 wt.% BMA emulsion at 40 ◦C for 15 min, while for HMA grafting, a higher temperature of 60 ◦C and longer grafting time of 30 min were needed. We also carried out BMA grafting at 60 ◦C. However, the grafting rate was too fast to control the graft yielding. For monomers with longer alkyl chains, preirradiation doses higher than 100 kGy were used to generate more radicals in the fabrics. This is because the steric hindrance effects of the monomers inhibited the graft polymerization from reaching a high degree of grafting. Furthermore, a mixture solvent of methanol and water in the ratio of 1:1 was used for OMA grafting. Here, the addition of methanol to the monomer solution increased the affinity between the fabric and the monomer, thereby enhancing the radiation grafting [53].

The loading of EHEP onto the grafted fabric was achieved by dropping the EHEP solution of ethanol onto the grafted fabric to reach a loading density of approximately 1.2 mmol/g. After removing ethanol by evaporation, the adsorbent was obtained.

#### *3.2. Materials Characterization*

The FTIR results shown in Figure 2 confirmed that the BMA, HMA, DMA, and OMA monomers were graft polymerized onto the PE/PP nonwoven fabrics and EHEP was loaded onto the OMA-grafted fabric. The peaks of the PE/PP nonwoven fabric only appeared at 1472, 1462, 1375, 731, and 718 cm−<sup>1</sup> , corresponding to the characteristic absorptions of PE [54], indicating that the PP fiber was completely coated by PE. After grafting, new peaks at 1730 and 1155 cm−<sup>1</sup> , attributed to the C=O and C–O stretching of methacrylate, respectively, were observed (Figure 2b–e) [55,56]. After loading EHEP onto the OMA-grafted fabric, new peaks at 1250 (P–O–C), 1050 (P–O–C), and 980 (P=O) cm−<sup>1</sup> were observed, as shown in Figure 2f [25]. These results indicated that the methacrylate monomers were grafted onto the fabrics and EHEP was loaded onto the OMA-grafted fabric. − − −

**Figure 2.** FTIR spectra of (**a**) PE/PP nonwoven fabric, (**b**) BMA-grafted PE/PP nonwoven fabric, (**c**) HMA-grafted PE/PP nonwoven fabric, (**d**) DMA-grafted PE/PP nonwoven fabric, (**e**) OMA-grafted PE/PP nonwoven fabric, and (**f**) OMA-adsorbent prepared by loading of EHEP onto the OMA-grafted PE/PP nonwoven fabric.

The surface properties of the BMA-, HMA-, DMA-, and OMA-grafted fabrics were evaluated by a contact angle meter. A high contact angle indicates the high hydrophobicity of the sample. Pictures of water droplets on the surface with the smallest and largest contact angles are shown in Figure 3a, b, respectively. The contact angle of the BMA-grafted fabric was 97◦ (Figure 3a), while that of the OMA-grafted fabric was 112◦ (Figure 3b). For comparison, the contact angles of the grafted fabrics are summarized in Figure 3c. The contact angle increased with the increase of the alkyl chain length of the grafted monomers. The OMA-grafted fabric had the highest hydrophobicity due to the longest alkyl chains of the grafted monomers as well as the highest degree of grafting (see Table 2). It was expected that the grafted fabric with high hydrophobicity was more conducive to the physical bonding of the alkyl chain of EAEH for loading.

**Figure 3.** Water droplets on the BMA-adsorbent (**a**) and OMA-adsorbent (**b**). Contact angle of the water droplets on the four fabric adsorbents (**c**).

#### *3.3. Batch Adsorption Tests*

The adsorbent performance was evaluated in advance by a batch adsorption test. The aqueous solution of 100 ppm Dy (III) and 100 ppm Nd (III) at pH 2 was used as the adsorption solution. After adsorption, the adsorbent was immersed in 1.0 M HCl solution to completely elute the adsorbed ions and washed with adequate water to conduct the adsorption test again.

The results of the batch adsorption test are shown in Table 3 and Figure 4. In the first adsorption test, all adsorbents had similar Dy (III) adsorptivity around 25.0 mg/g. The EHEP loaded in the fabric is a cationic extractant, which is known to extract metal ions from aqueous solution and can be labeled HA. The adsorption is an ion-exchange process in which one Dy (III) ion combines three EHEPs to form a DyA<sup>3</sup> structure in the adsorbent [57,58]. The similar adsorptivity was due to the similar EHEP loading (1.20 mmol/g) of the four adsorbents. However, the Nd (III) adsorptivity for each adsorbent was considerably small (less than 1.0 mg/g). Therefore, the EHEP-loaded adsorbents had a high adsorption selectivity for Dy (III) and could be used for separation and recovery.


**Table 3.** Summary of the first and repeated batch adsorption tests.

\* First adsorption was performed using the new adsorbent, and W<sup>b</sup> is the dry weight of the new adsorbent, CDy-1 and CNd-1 are the Dy(III) and Nd(III) adsorptivities of the first adsorption, respectively; \*\* Repeated adsorption was performed after the adsorbent diluted and adequate water-washed, CDy-r and CNd-r are the Dy(III) and Nd(III) adsorptivities of the repeated adsorption, respectively, and W<sup>a</sup> is the dry weight of the used adsorbent after the repeated adsorption and dilution.

**Figure 4.** The Dy (III) and Nd (III) adsorptivity of the four nonwoven fabric adsorbents. (**a**) First adsorption test. (**b**) Second adsorption test using the refreshed adsorbents after elution and washing. Initial adsorption solution: 100 ppm Dy (III) and 100 ppm Nd (III) at a pH of 2.0 at 25 ◦C.

In the repeated adsorption tests, the OMA-adsorbent retained a high Dy (III) adsorptivity of 25.3 mg/g. In contrast, the Dy (III) adsorptivities of BMA-, HMA-, and DMA-adsorbents were significantly reduced to were 11.4, 15.0, and 22.7 mg/g, respectively. The decrease of Dy (III) adsorptivity might be due to the loss of EHEP loaded in the fabric during the repeated tests. As shown in Table 3, after repeated adsorption tests, the weight of the OMA-adsorbent was almost unchanged, while the weight of the BMA-adsorbent was reduced by 26%. The shorter the alkyl chain length of the grafted monomer, the more the weight of the adsorbent decreased due to the loss of EHEP. According to these results, the OMA-adsorbent with the longest alkyl chain was chosen for the column adsorption test.

#### *3.4. Column Adsorption Tests*

Column adsorption and elution were carried out using the same adsorption and elution solutions as the above batch tests. The solution was passed through the column at a space velocity of 100 h−<sup>1</sup> . As shown in Figure 5, the Dy (III) was completely adsorbed up to a higher bed volume (BV) of 80. After that, the concentration of Dy (III) in the effluent gradually increased, reaching 98 ppm at a BV of 400 (similar to the concentration of the fed solution, 100 ppm). The total Dy (III) adsorbed from the solution was calculated using Equation (5) to be 43.6 mg/g. The adsorption is an ion-exchange process between the metal ions and the proton of EHEP loaded in the fabrics; that is, one Dy (III) ion can bond with three phosphate groups. Therefore, for a 1.2 mmol/g EHEP-loaded adsorbent, the calculated adsorption capacity is close to 64.8 mg/g. The detected value of 43.6 mg/g is lower than the calculated value, which is due to the adsorption equilibria at the low Dy (III) concentration of the feed solution. Even then, it is still much higher than in the case of using hybrid silica nanoparticles, as reported by Topel et al., where the Dy (III) adsorption is 0.019 mmol/g or 30.9 mg/g [57]. In contrast, the Nd (III) was completely adsorbed up to a lower BV of 40, and the Nd (III) concentration rapidly increased up to 130 ppm at a BV of 144, which was higher than that of the fed solution (100 ppm). This is because the adsorbed Nd (III) was replaced by Dy (III), indicating that the OMA-adsorbent was favorable for Dy (III) adsorption. The Nd (III) adsorptivity of the adsorbents in the column was also calculated using Equation (5) to be 4.2 mg/g, which was one-tenth of the Dy (III) adsorption.

The adsorbed Dy (III) and Nd (III) were eluted by passing 1.0 M HCl solution through the column. The maximum concentrations of Dy (III) and Nd (III) in the effluent were 373 and 38 ppm, respectively. The recovery ratios of Dy (III) and Nd (III) calculated using Equations (6) and (7) were 99% and 98%, respectively, indicating that almost all metal ions were eluted by the 1.0 M HCl solution within a BV of 160 (from 550 to 710 BV in Figure 5).

After the first adsorption, water washing, HCl elution, and water washing, the repeated column test was continued (Figure 5). The concentration curves of Dy (III) and Nd (III) for the repeated adsorption test show similar behavior as the first adsorption test. These results indicate that the OMA-adsorbent was stable for repeated use in the separation and recovery of Dy (III) and Nd (III) ions from an aqueous solution.

**Figure 5.** Profile of the adsorption and elution of Dy (III) and Nd (III) ions with an OMA-adsorbent. Adsorption solution: 100 ppm Dy (III) and 100 ppm Nd (III), pH 2.0; elution solution: 1.0 M HCl; space velocity (SV) = 100 h−<sup>1</sup> ; temperature = 25 ◦C; the total 1200 BV means that the adsorption–elution– adsorption process was operated for 12 h under the fixed space velocity of 100 h−<sup>1</sup> .

#### **4. Conclusions**

A fabric adsorbent for the separation and recovery of Dy (III) and Nd (III) from an aqueous solution was successfully prepared by graft polymerization of methacrylate with a long alkyl chain onto the nonwoven fabric and loading EHEP by hydrophobic interaction and chain entanglement between the alkyl chains.

In the batch adsorption tests, the adsorbents showed a high Dy (III) adsorptivity above 25.0 mg/g and a low Nd (III) adsorptivity below 1.0 mg/g, indicating that the adsorbents had a high Dy (III) selective adsorption. However, only the OMA-adsorbent with the longest alkyl chain was stable and retained its high Dy (III) adsorption performance in repeated adsorption tests.

In the column adsorption test with the OMA-adsorbent, the adsorptivities of Dy (III) and Nd (III) were 43.6 and 4.2 mg/g, respectively. The Dy (III) adsorptivity was approximately ten times higher than that of the Nd (III) adsorptivity. Similar adsorption performance of the adsorbents was observed in the repeated tests. These results demonstrate that the OMA-adsorbent was stable for repeated use. The high stability of the OMA-adsorbents due to the loss of EHEP was suppressed by the strong hydrophobic interaction and chain entanglement between the long alkyl chains.

The OMA-adsorbent can be synthesized easily and economically by immersing the irradiated nonwoven fabric in the monomer solution and EHEP solution in sequence. The obtained adsorbent can be used in batch mode or column mode without any other separation process. Even if the adsorbent is operated in a strong acid, it is stable without any weight loss. Furthermore, the adsorbent has a high selectivity to Dy (III) ions. Therefore, the OMA-adsorbent developed in this study can effectively separate and recover Dy (III) and Nd (III) from an aqueous solution and is expected to contribute to the recovery of rare-earth metals from NdFeB permanent magnet scraps in the future.

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

**Funding:** This research was partly funded by the Japan Society for the Promotion of Science (JSPS) KAKENHI Grant Numbers JP17K00632 (Grant-in-Aid for Scientific Research (C)).

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

#### **References**


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

© 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*
