Fast Procedure for Removing Silver Species in Waters Using a Simple Magnetic Nanomaterial

: The increase in the production and use of disinfectants containing silver atoms (in both its ionic and nanomeric forms) in their formulation, due to the global pandemic situation caused by COVID-19, has increased the presence of silver species in wastewater. Moreover, silver atoms are now considered as emerging pollutants in water. In this work, we propose a novel method for the instantaneous and simultaneous removal of ionic and nanomeric silver in water samples, using a previously unpublished methodology consisting of the in situ formation of magnetic nanoparticles in the aqueous samples to be treated. While the nanoparticle precursors react to form them, the silver atoms present in the sample are adsorbed onto them due to a strong electrostatic interaction. As the ﬁnal nanoparticles are magnetic, they can be easily removed from the aqueous medium using a magnet, leaving the samples free of silver species. The innovative feature of the method is that the adsorbent is synthesized in situ, within the sample to be treated, making the approach a low-cost, easy-to-perform solution. Temperature, contact time, dose of Fe 3 O 4 , and concentration of nanomeric and ionic silver were investigated. The results showed that at 50 ◦ C, 100% of both silver species were removed from the water samples simultaneously. The surface of Fe 3 O 4 was characterized before and after the application of the removal process using energy-dispersive X-ray spectroscopy and Field Emission scanning electron microscopy. Adsorption kinetics and equilibrium isotherms studied reveal a Langmuir-type physicochemical process. The procedure has been applied to different water samples (river and drinking water) with excellent results, making the method a new standard for the removal of ionic and nanomeric silver. In addition, the nanoparticles formed could be recycled and reused for other analytical and decontamination purposes.


Introduction
Emerging contaminants are species present in the environment that are not recognized as contaminants but whose presence is of emerging concern because of their potential environmental and health effects [1]. Although many of them are found in very low concentrations, they cause serious problems in ecosystems, especially when they are discharged into aquatic environments. To detect and quantify these contaminants, it has been necessary to develop new analytical techniques that are sufficiently sensitive due to the low concentration of some of these emerging pollutants found in the environment [2,3]. Some of the substances considered as emerging pollutants in water include pharmaceuticals, detergents, dyes, and personal care products. Recently, the widespread use of nanomaterials has also contributed to the identification of metallic nanoparticles as emerging contaminants [4].
Currently, many products contain ionic and nanomeric silver in their formulation due to the well-known disinfectant and antifungal properties of silver [5]. Nanomeric silvers Field Emission scanning electron microscopy (FESEM) was performed using a Zeiss Merlin VP Compact microscope (Oberkochen, Germany). The energy-dispersive X-ray (EDX) detector used was a Bruker Quantax 400 (Vienna, Austria). Samples were prepared by depositing suspension drops on a copper transmission electron microscopy grid (covered with a perforated carbon film) and air-dried. Experimental conditions were high vacuum Separations 2023, 10,398 4 of 17 at 2 kV for detection of secondary and backscattered electrons, and 15 kV and 5 kV for EDX spectra and mapping analysis, respectively.
Nd-Fe-B permanent magnets were supplied (Supermagnete, Webcraft GmbH, Gottmadingen, Germany) as small blocks and discs providing 33 and 38 kgf, respectively, for separating the magnetic particles. Particle size and zeta potential data were obtained with a Malvern Zetasizer nano zsp instrument (Malvern, UK).

In Situ Preparation of Fe 3 O 4 Nanoparticles and Silver Removal Procedure
A 150 µL of FeCl 3 ·6H 2 O (0.1 M) solution and a 150 µL FeCl 2 ·4H 2 O (0.2 M) solution were added to 10 mL of a water solution at individual concentrations of silver species of 100 µg L -1 . Subsequently, the mixture was heated at 50 • C for five minutes. Then, 50 µL of a concentrated ammonia solution was added to give the solution a basic pH, necessary for the formation of nanoparticles according to the chemical reaction that takes place for the formation of Fe 3 O 4 , as shown below [10]: The Fe 3 O 4 nanoparticles were rapidly synthesized in the solution to be treated and, during their formation process, the silver atoms were eliminated from the aqueous solution by means of a mixed process based on the adsorption of silver on the ferrite nanoparticles and the ability of the iron oxide to collect metals without excluding electrostatic interactions between the silver and iron atoms.
After the nanoparticles were removed from the medium with a neodymium magnet, the silver-free (ionic and nanomeric) aqueous solution was decanted and analyzed by Electrothermal Atomic Absorption Spectrometry (ETAAS), confirming that the water sample was free of silver.

Characterization of Fe 3 O 4
The basic characterization of the magnetic material obtained by the procedure described in Section 2.2 has already been described [61][62][63]. Figure 1 shows a Field Emission Scanning Electron Microscopy (FESEM) image of the surface of Fe 3 O 4 formed in situ when the aqueous solution contains no silver atoms. Figure 2 shows the energy-dispersive X-ray (EDX) spectrum corresponding to the whole area of the image shown in Figure 1, together with the atomic concentration table corresponding to Fe 3 O 4 , which shows strong peak signals for O and Fe, as expected. The peaks for Cu, C, Si and Al correspond to the grid and sample holder used in the FESEM.     Detection of silver afte r adsorption on Fe3O4 by standard FESEM analysis was challenging due to the very low concentration of silver in the samples. However, backscattered electron (BSE) analysis was able to distinguish Ag in the samples, with these atoms appearing as brighter structures due to their higher atomic number [64,65]. Accordingly, Figure 3 shows BSE images for Fe3O4 after the removal of Ag, where the brighter areas Detection of silver after adsorption on Fe 3 O 4 by standard FESEM analysis was challenging due to the very low concentration of silver in the samples. However, backscattered electron (BSE) analysis was able to distinguish Ag in the samples, with these atoms appearing as brighter structures due to their higher atomic number [64,65]. Accordingly, Figure 3 shows BSE images for Fe 3 O 4 after the removal of Ag, where the brighter areas correspond to Ag, although it is not possible to distinguish between the silver species. Figure 4 shows the EDX spectrum and atomic concentration table corresponding to the whole area of Figure 3. As mentioned above, the concentration of silver is very low, but the EDX system automatically detects the presence of Ag. correspond to Ag, although it is not possible to distinguish between the silver species. Figure 4 shows the EDX spectrum and atomic concentration table corresponding to the whole area of Figure 3. As mentioned above, the concentration of silver is very low, but the EDX system automatically detects the presence of Ag.

Effect of Fe 3 O 4 Precursors Volume
The formation of ferrite nanoparticles (Fe 3 O 4 ) used in the process results from the reaction of Fe 3+ and Fe 2+ in a basic medium. To optimize the formation process, the dosage of Fe 3 O 4 precursors was studied at room temperature. To develop this study, the volume of aqueous solutions of FeCl 3 ·6H 2 O (0.1 M) and FeCl 2 ·4H 2 O (0.2 M) was varied from 50 µL to 200 µL. The silver elimination process described above was then applied individually to each Ag species to show that they exhibited similar behaviour. Thus, 10 mL of two different aqueous solutions containing concentrations of ionic and nanomeric silver of 100 µg L −1 were tested. Figure 5 shows that the best results in terms of removal efficiency were obtained when 150 µL of each precursor solution was used for both Ag(0) (curve a) and Ag(I) (curve b). This result remains constant even when 200 µL of each precursor is used.

Effect of Contact Time
The contact time between water solutions (containing isolated species Ag(0) and Ag(I), and for a solution containing both species) and Fe 3 O 4 was studied in order to achieve maximum removal efficiency. Experiments were carried out for silver concentrations equal to 100 µg L −1 , T = 50 • C, and 150 µL of solutions of FeCl 3 ·6H 2 O (0.1 M) and FeCl 2 ·4H 2 O (0.2 M). After the addition of 50 µL of 25% ammonia solution, 500 µL aliquots were taken at different times. The magnet was immediately applied, and the aqueous phase was separated to measure the silver content. The results obtained ( Figure 6) showed a constant maximum removal efficiency after 5 min. of aqueous solutions of FeCl3·6H2O (0.1 M) and FeCl2·4H2O (0.2 M) was varied from 50 µL to 200 µL. The silver elimination process described above was then applied individually to each Ag species to show that they exhibited similar behaviour. Thus, 10 mL of two different aqueous solutions containing concentrations of ionic and nanomeric silver of 100 µg L −1 were tested. Figure 5 shows that the best results in terms of removal efficiency were obtained when 150 µL of each precursor solution was used for both Ag(0) (curve a) and Ag(I) (curve b). This result remains constant even when 200 µL of each precursor is used.

Effect of Contact Time
The contact time between water solutions (containing isolated species Ag(0) and Ag(I), and for a solution containing both species) and Fe3O4 was studied in order to achieve maximum removal efficiency. Experiments were carried out for silver concentrations equal to 100 µg L −1 , T = 50 °C, and 150 µL of solutions of FeCl3·6H2O (0.1 M) and FeCl2·4H2O (0.2 M). After the addition of 50 µL of 25% ammonia solution, 500 µL aliquots were taken at different times. The magnet was immediately applied, and the aqueous phase was separated to measure the silver content. The results obtained ( Figure 6) showed a constant maximum removal efficiency after 5 min. Pseudo-first-order and pseudo-second-order kinetic models were used to study the rate of sorption. The pseudo-first-order model allows us to check whether the rate of sorption point occupancy is proportional to the number of unoccupied sites. The pseudo-second-order kinetic model allows us to test whether chemisorption is the step that controls the kinetics of the process. The expressions used for these two models were are the adsorption rate constants and t q and q (mg·g −1 ) are the amount of metal adsorbed at each time and at equilibrium, respec- Pseudo-first-order and pseudo-second-order kinetic models were used to study the rate of sorption. The pseudo-first-order model allows us to check whether the rate of sorption point occupancy is proportional to the number of unoccupied sites. The pseudosecond-order kinetic model allows us to test whether chemisorption is the step that controls the kinetics of the process. The expressions used for these two models were where k 1 (min −1 ) and k 2 (g·mg −1 ·min −1 ) are the adsorption rate constants and q t and q e (mg·g −1 ) are the amount of metal adsorbed at each time and at equilibrium, respectively [66,67]. The adsorption kinetics of Ag(0) and Ag(I) adsorption following these two models were carried out on aqueous samples containing 100 µg L −1 of both ions. At different times, an aliquot of 500 µL was taken from the medium, and the magnet was immediately applied to separate the adsorbent phase. The content of silver was measured by ETAAS. The obtained results are shown in Figure 7. The kinetic parameters and correlation coefficients are given in Table 2. As can be seen, both results fit better with a pseudo-second-order model, which seems to indicate a chemisorption process.

Study of the Effect of the Temperature and Adsorption Isotherms
Adsorption isotherms describe the dependence of the equilibrium adsorption capacity, (mg g −1 ), and the equilibrium concentration of the adsorbate, (mg L −1 ) [63]. The study was conducted for the simultaneous removal of silver (ionic and nanomeric) at concentrations equal to 50, 100, 150, and 200 µg L −1 and at temperatures of 20, 30, 50, and 70 °C. Contact time was instantaneous, pH was set to basic, and the maximum removal efficiency of 50 °C was achieved, staying constant up to 70 °C.
The adsorption isotherm models considered were Langmuir, which takes into account homogeneous monolayer adsorption, and Freundlich, which takes into account multilayer adsorption and affinity between adsorption surface sites [68]. These models gave the best fit (adjusted R 2 > 0.99) compared to other models such as Temkin.

Study of the Effect of the Temperature and Adsorption Isotherms
Adsorption isotherms describe the dependence of the equilibrium adsorption capacity, q e (mg g −1 ), and the equilibrium concentration of the adsorbate, C e (mg L −1 ) [63]. The study was conducted for the simultaneous removal of silver (ionic and nanomeric) at concentrations C e equal to 50, 100, 150, and 200 µg L −1 and at temperatures of 20, 30, 50, and 70 • C. Contact time was instantaneous, pH was set to basic, and the maximum removal efficiency of 50 • C was achieved, staying constant up to 70 • C. The adsorption isotherm models considered were Langmuir, which takes into account homogeneous monolayer adsorption, and Freundlich, which takes into account multilayer adsorption and affinity between adsorption surface sites [68]. These models gave the best fit (adjusted R 2 > 0.99) compared to other models such as Temkin.
For the Langmuir model, K L (L mg −1 ) is the Langmuir adsorption constant and q m (mg g −1 ) is the maximum adsorption capacity. For the Freundlich model, K F and 1/n and are the adsorption capacity and adsorption intensity constants, respectively.
The fitting of the theoretical models in Equation (1) to the data is summarized in Table 2. Fits corresponding to isotherm at T = 70 • C are not shown because the isotherm matches with that of T = 50 • C, both achieving 100% removal efficiency for all concentrations. Quantities computed in software Origin 2022 were reduced R 2 and χ 2 , as well as Bayesian and Akaike information criteria, in order to compare the isotherm models. As depicted in Table 3, the Langmuir equation gives rise to the best results, maximum R 2 and minimum χ 2 , for all temperatures. Additionally, Bayesian and Akaike criteria indicate that the Langmuir model is more likely to describe the equilibrium adsorption of silver onto Fe 3 O 4 . Graphical results for the Langmuir model are shown in Figure 8. Table 3. Nonlinear fits of Langmuir and Freundlich models (Equation (1)) to experimental data presented in Figure 6. The Langmuir model gives rise to the best fitting results for all temperatures. For the Langmuir model, KL (L mg −1 ) is the Langmuir adsorption constant and qm (mg g −1 ) is the maximum adsorption capacity. For the Freundlich model, KF and 1/n and are the adsorption capacity and adsorption intensity constants, respectively.

Langmuir
The fitting of the theoretical models in Equation (1) to the data is summarized in Table  2. Fits corresponding to isotherm at = 70 °C are not shown because the isotherm matches with that of = 50 °C, both achieving 100% removal efficiency for all concentrations. Quantities computed in software Origin 2022 were reduced and , as well as Bayesian and Akaike information criteria, in order to compare the isotherm models. As depicted in Table 3, the Langmuir equation gives rise to the best results, maximum and minimum , for all temperatures. Additionally, Bayesian and Akaike criteria indicate that the Langmuir model is more likely to describe the equilibrium adsorption of silver onto Fe3O4. Graphical results for the Langmuir model are shown in Figure 8.   (1)) to experimental data presented in Figure 6. The Langmuir model gives rise to the best fitting results for all temperatures. For the removal of Ag(I) and Ag(0), the adjustment of each Langmuir isotherm gives a maximum adsorption capacity q m of 142.3 and 132.1 mg g −1 , respectively, at 50 • C.

Langmuir Model Freundlich Model
The analysis of the standard Gibbs free energy ∆G 0 (kJ mol −1 ) brings important information about the nature of the adsorption process, identifying it as chemisorption, physisorption, or a mixture of both [63]. Negative ∆G 0 values correspond to spontaneous adsorption, chemisorption lying close to the range [−400, −80] kJ mol −1 , and physisorption to [−20, 0] kJ mol −1 . For each temperature, the ∆G 0 value is determined as follows: where T is the absolute temperature, R is the gas constant, and K t is the equilibrium thermodynamic constant, whose value should be appropriately determined at each T from the analysis of the equilibrium isotherms, transforming K L into the adimensional quantity K t [69]: In this expression, [adsorbate] 0 = 1 mol L −1 represents the standard concentration of the adsorbate, M is the molecular mass of the adsorbate (g mol −1 ), and γ = 1 accounts for the activity coefficient of the dilute solution. The results for temperatures 24, 30, 50, and 70 • C show that ∆G 0 values lie within the range [−34.13, −28.28], which can be considered as a mixture between physical and chemical adsorption, a phenomenon previously reported for the adsorption of metal ions by employing adsorbents based on magnetic nanoparticles [40].

Mechanism of Adsorption
The zeta potential represents the charge of a nanoparticle relative to the surrounding conditions. However, the zeta potential is not a true measure of the individual molecular surface charge; rather, it is a measure of the electrical double layer created by the surrounding ions (i.e., counterions) in solution. These counterions play a role in the calculation of the zeta potential measurement. All particle systems in an aqueous medium carry an electrical charge, which can be positive, negative, or neutral. Nanoparticles functionalized on the surface with, e.g., carboxylic acids will produce a negatively charged surface due to the dissociation of an acidic group, whereas the dissociation of a basic group on the surface of a nanoparticle will produce a positively charged surface. For unmodified nanoparticles, the individual atoms that make up the surface of the particle determine its charge. Nanoparticle agglomeration occurs when individual particles are held together by weak interparticle interactions, including solvation forces, van der Waals forces, electrostatic attraction, and hydrophobic interactions [70][71][72]. In most cases, the agglomerated state is reversible, but only if additional entropy (e.g., sonication or homogenization) or ions (changing H + ) are added to the system. It is suggested that this method could be used to select the size of specific nanopopulations within a particle suspension.
In aqueous systems, the surface of iron oxides is covered with Fe-OH groups [61]. Other groups such as FeOH 2 + and FeO − are present on the surface of magnetic ferrite at pH values below and above pH pzc , respectively. The zero charge potential of magnetite (pH pzc ) is the pH at which the surface concentrations of FeOH 2 + and FeO − are equal. Electrostatic forces between the species in solution and the charged surface are responsible for the adsorption. The measured pH pzc of magnetite was 7.5, although values ranging from 5 to 7.5 have been recorded [73]. The zeta potential measured on magnetic ferrite was 9.08 mV at zero charge potential, suggesting that positively charged particles may be retained on the surface by electrostatic attraction. As pH increases, the FeOH 2 + groups are converted to FeO − , increasing the net negative surface charge and thus the interaction with positive species [62]: In the case of silver particles, the surface charge density depends on the method of synthesis and the presence or absence of functionalizing agents that modify the surface of the nanoparticles [74,75]. In the case of AgNPs obtained by using NaBH 4 and silver nitrate and finally stabilized in citrate buffer, the pH pzc is 3.5 [76] and 3 [77], respectively. These values indicate that at pH above 3-3.5, the silver nanoparticles are negatively charged. Consequently, the electrostatic attraction and perhaps coordination adsorption between the ferrite and the silver nanoparticles are the removal mechanisms.
The Ag(I) ion is stable as such in aqueous solutions up to slightly basic pH values, where it starts to evolve into Ag(OH) hydroxide and Ag(OH) 2 − anion [78]. Therefore, at pH = 8, it is still present in the aqueous solution as a positive species and can be retained on the ferrite, which is already negatively charged at this pH: The same applies to the Ag(0) species. However, when working with previously synthesized ferrite, the retention of Ag(0) is not as effective as when the ferrite is synthesized in situ. Therefore, in addition to an electrostatic interaction mechanism, some other interaction process with the iron atoms of the ferrite structure must be at work during formation.

Comparison with Other Removal Methods
The evaluation of the efficacy of conventional water treatment processes (i.e., coagulation, flocculation, sedimentation and sand filtration) in removing stabilized Ag NPs from natural water has been studied [75]. Flocculation and settling were found to be key steps for the removal of silver nanoparticles.
The simplicity of operation and cost-effectiveness have led to the investigation of suitable sustainable alternatives such as adsorption. Several studies have been carried out on different adsorbents such as activated carbons, clays, biowaste materials, cellulosic materials, zeolites, graphene, biochars, and synthetic materials. Table 4 shows several proposed materials for silver removal from water samples. As can be seen, the adsorption capacity of ferrite synthesized in situ is higher than that of most of the adsorbents listed in Table 4. The removal of silver nanoparticles is completed in only 5 min. The process of material synthesis is very simple. In addition, the amount of adsorbent material required is very small. Therefore, we can say that the proposed material is a suitable candidate for the removal of heavy metals.

Study of Competition with Other Ions Present in Water. Application to Real Water Samples and Reuse of Adsorbent
As there are many ions present in any type of wastewater, the possibility that the presence of other ions commonly found in water could interfere with the silver removal process was investigated.
To check the possible competition, the approach proposed in this work was carried out by placing the medium-high concentrations (500 and 1000 mg L -1 ) of Ca 2+ , Pb 2+ , Hg 2+ , As 3+ , As 5+ , Mg 2+ , Cl -, NO 3 -, and SO 4 2-. In every case, the removal of silver species was not affected by the presence of these ions when the concentrations were 500 µg L -1 . When the concentrations of ions were 1000 µg L -1 , the silver elimination process was affected exclusively in presence of Hg. In this case, the silver removal efficiency dropped to 92%, which is still a high percentage. In addition, Hg 2+ is not a species that is commonly found at such a high concentration in any type of water. The proposed procedure was applied to real river water samples from the surrounding area at different points and to drinking water samples. The results (see Table 5) showed that the method was fully applicable to these water samples, as silver removal efficiencies ranging from 97.1 to 99.2% were achieved in all cases. As the concentration of total silver in the water samples was below the limit of detection (LOD), the samples were spiked with a known concentration of ionic and nanomeric silver. Experiments were carried out to verify if the proposed procedure is adequate when the levels of ionic silver and nanomeric silver are less than 10 µg L −1 . At these levels of silver concentration, it is not possible to directly measure aqueous samples because of the sensitivity of the detection technique. In these cases, we have resorted to measuring the concentration of silver retained from its desorption in an acid medium containing thiourea [5]. This desorption is carried out in a small volume (100 µL), which makes it possible to reach an enrichment factor close to 100. The results obtained indicate quantitative retentions even at levels of 0.3 µg L −1 of silver in aqueous samples.
At present, the use of silver-functionalized magnetic nanoparticles is of great interest in many scientific fields, whether for use as extraction media in analytical chemistry [51] or as an adsorbent medium for other species [3,50]. Consequently, in this work, we offered a double solution: on the one hand, the removal of silver from a contaminated medium and, on the other hand, the reuse of the adsorbent containing the silver for other purposes, thus establishing a circular process.
The adsorption of silver species occurs during ferrite formation in the proposed process. The process is therefore instantaneous and quantitative. Therefore, it does not make sense to study the reuse of a product that has already been synthesized. However, we have studied the possibility of recovering the magnetic material after e desorption of the retained species. To study the possibility of reusing the magnetic material formed in situ, silver was desorbed by adding 100 µL of 0.1 M nitric acid, and the mixture underwent ultrasound for 3 min. Afterwards, the ferrite was isolated by the approach of the magnet and was clean and free of silver, so it could be used for other purposes.
In this regard, we have tested the validity of recycled material for the retention of amoxicillin and ibuprofen. In fact, in some previous works of our research group, several methods to remove some emerging contaminants such as amoxicillin and ibuprofen were developed [3,63]. Here, magnetic ferrite nanoparticles functionalized with nanomeric silver were used as sorbents. Therefore, the adsorbent material used in these previous works corresponds to the one obtained in this work after the removal of silver from water. For this reason, the resulting material after silver removal was added to aqueous solutions containing amoxicillin and ibuprofen, and the optimal experimental conditions described in these previous works were applied. The result was that both contaminants were completely removed from the aqueous medium. Therefore, after silver from water is removed from water, the resulting material can be successfully used in other processes to remove emerging contaminants such as amoxicillin and ibuprofen.

Conclusions
A novel method for the total and instantaneous removal of ionic and nanomeric silver via the in situ formation of magnetic nanoparticles has been introduced. The extraordinary innovation of synthesizing the adsorbent in solution allows the process to remove a high concentration of silver from water samples in just a few seconds, achieving 100% removal efficiency. It is a cost-effective and simple process. In addition, the approach can be applied to real water samples with excellent results. Another important advantage is the possibility of reusing the silver-containing adsorbent for other analytical purposes, as silver-functionalized Fe 3 O 4 is of great scientific interest today. In addition, adsorption isotherms and temperature studies show a mixture of physical and chemical adsorption, characterized by Langmuir equilibrium isotherms.
Therefore, this novel method is postulated as an efficient alternative to other procedures currently available in the literature, which are considered more tedious, less rapid, and more expensive, and which are also unable to achieve 100% removal efficiency for ionic and nanomeric silver.

Acknowledgments:
The authors want to thank the University Centre of Defence at the Spanish Air Force Academy, MDE-UPCT.

Conflicts of Interest:
The authors declare that the work described has not been published previously, that it is not under consideration for publication elsewhere and that this publication is approved by all authors.