Protection of Thiol Groups on the Surface of Magnetic Adsorbents and Their Application for Wastewater Treatment

The magnetite nanoparticles were functionalized with silica shells bearing mercaptopropyl (monofunctional) and mercaptopropyl-and-alkyl groups (bifunctional) by single-step sol-gel technique. The influence of synthetic conditions leading to increased amounts of active functional groups on the surface and improved capacity in the uptake of Ag(I), Cd(II), Hg(II), and Pb(II) cations was revealed. The physicochemical properties of obtained magnetic nanocomposites were investigated by FTIR, Raman, XRD, TEM, SEM, low-temperature nitrogen ad-/desorption measurements, TGA, and chemical microanalysis highlighting the efficiency of functionalization and mechanisms of the preparation procedures. The removal of the main group of heavy metal cations was studied in dependence from the pH, contact time and equilibrium concentration to analyze the complexes composition for the large scale production of improved adsorbents. It was demonstrated that introduction of the alkyl groups into the surface layer prevents the formation of the disulfide bonds between adjacent thiol groups. The obtained adsorbents were employed to treat real wastewater from Ruskov, Slovakia with concentration of Fe 319 ng/cm3, Cu 23.7 ng/cm3, Zn 36 ng/cm3, Mn 503 ng/cm3, Al 21 ng/cm3, As 34 ng/cm3, Pb 5.8 ng/cm3, Ni 35 ng/cm3, Co 4.2 ng/cm3, Cr 9.4 ng/cm3, Sb 6 ng/cm3, Cd 5 ng/cm3. These materials proved to be highly effective in the removal of 50% of all metal ions, espeсially Zn, Cd, and Pb ions from it and turned recyclable, opening for their sustainable use in water purification.

Heavy metals are a group of elements (with a molecular mass above 50) that actively participate in biological processes forming part of many enzymes, but they have a harmful effect on the body causing poisoning and mutations at a certain concentration. These include: lead, zinc, cadmium, mercury, molybdenum, chromium, manganese, nickel, tin, cobalt, titanium, copper, vanadium. For example, сhromium (in its hexavalent form) and arsenic are carcinogens; cadmium causes a degenerative bone disease; and mercury and lead damage the central nervous system. Other heavy metals noted for their potentially hazardous nature, usually as toxic environmental pollutants, include manganese (central nervous system damage); cobalt and nickel (carcinogens); copper, zinc, selenium and silver (endocrine disruption, congenital disorders, or general toxic effects in fish, plants, birds, or other aquatic organisms).
Sources of heavy metals are divided into natural (weathering of rocks and minerals, erosion processes, volcanic activity) and man-made (mining and processing of minerals, fuel combustion, traffic, agricultural activities). There are many methods of water treatment, including mechanical, biological, physical and chemical, but adsorption is very important for removing microquantities of metal ions that remain after previous purifications. For this purpose, surface of inorganic adsorbents are modified by various complex and chelate groups [1][2][3][4][5]

Results and Discussion
Synthesis and content of Thiol and Alkyl groups. Procedures, where the sol created by TEOS (tetraethyl orthosilicate) and MPTMS (and MTES (methyltriethoxysilane) or PTES (n-propyltriethoxysilane)) was added dropwise to the suspension of magnetite during 20 minutes were applied. In order to form porous silica shell over magnetite praticles, the ammonium fluoride was emloyed as catalyst 29 in the synthesis of samples S1, SM1 and SP1. The rest of saples S2, SM2, SP2 and S3 were synthetised without catalyst to compare physical charcteristics of obtained materials. The composition of the reaction mixture and content of functional groups present on the magnetic particles functionalized by 3-mercaptopropyl groups are given in Table 1. The molar ratio between Fe 3 O 4 and TEOS (1:20), which provided stability of obtained materials in an acidic environment, was the same as used previously 30 . Table 1 shows the content of introduced functional groups calculated from the data provided by different methods. The amount of 3-mercaptopropyl groups was calculated from the elemental analysis and EDXS data. It can be seen that EDXS data correlate within an error, thus the data of elemental analysis for sulfur were used in It was observed that the hydrolysis rate decreases with increasing length of alkyl groups connected to the silicon atom 31 . The same effect is observed for samples with 3-aminopropyl groups with simultaneous introduction of a methyl or n-propyl groups (bifunctional materials). Almost double increase in amount of 3-aminopropyl groups when compared with the theoretically calculated amounts can be observed as well 32 .
The amount of methyl and n-propyl groups in the obtained samples were evaluated based on elemental analysis (Table 1) and thermograms (Fig. FS1), where the weight loss of the part which corresponds to 3-mercaptopropyl groups (elemental analysis data for sulfur) was substracted from the total weight loss assuming that the decomposition of organic groups begins over 200 °C. Thus, the quantities of methyl (M) or n-propyl (P) groups were found to be 2.3 and 0.6 mmol/g in samples SM1 and SM2, respectively, and 2.3 and 0.3 mmol/g in samples SP1 and SP2, respectively. In addition, the amounts of alkyl groups were calculated from the elemental analysis of carbon (Table 1). It is obvious that the concentration obtained from these two methods correlate well. Smaller amount of both 3-mercaptopropyl and alkyl groups in the S2, SM2 and SP2 samples, compared to the S1, SM1 and SP1 samples, justified the importance of ammonium fluoride as catalyst. It can be explained that in the syntheses of the S1, SM1 and SP1 samples the hydrolysis and condensation of silanes proceeded to a major degree, consequently leading to formation of bigger amounts of functional groups on the material's surface.
IR and Raman spectra. The presence of both polysiloxane layer and the functional groups, on the surface of magnetite particles, was confirmed by IR spectroscopy (Fig. 2a). Strong absorption band in the 1000-1200 cm −1  Groups of bands in the DRIFT spectra indicated that functional groups were also present. The presence of propyl chains was obvious from a set of bands at 1350, 1411 and 2840-2980 cm −1 . These bands are assigned to ω(CH 2 ), δ(Si-CH 2 ) and ν s,аs (СН), respectively. In the DRIFT spectra of samples SM1 and SM2 the two low-intensity absorption bands at 1280 and 1437 cm −1 , assigned to δ s (CH 3 ) and δ аs (CH 3 ), respectively, were present. These bands verified the presence of methyl groups in the samples. A low-intensity absorption band at 2575 cm −1 was assigned to the ν(SH) stretching vibrations of 3-mercaptopropyl group. This band was absent in almost all spectra, or had a very low intensity, as well as in other reported studies dedicated to functionalization of magnetite by MPTMS. This can be attributed to relatively low concentration of functional groups present on the surface 9,10 . However, one should not exclude possibility of partial oxidation of the 3-mercapto groups to disulfide. Therefore, the Raman spectra were also recorded (Fig. 2b). In these spectra could be seen three lines at 509, 640 and 2575 cm −1 attributed to the stretching vibrations of S-S, C-S and S-H bonds, respectively 34,35 . It could be observed that all spectra contained lines attributed to C-S and S-S stretching vibrations. The latter confirms the presence of the disulfide bridges which apparently formed in the oxidation of 3-mercaptopropyl groups.
Only three of them, two bifunctional SM1 and SP1, and freshly synthesized sample S1, contained the line assigned to the S-H stretching vibrations. The intensity of this line also represents the number of functional groups per unit of the surface and correlates with the data presented in the Table 1. Thus, the introduction of alkyl groups to the surface layer of magnetically sensitive materials with 3-mercaptopropyl groups partially inhibits their oxidation.
Powder XRD. Figure 3 shows the XRD patterns of the functionalized samples. The reflections at 13  . Moreover, the thickness of the polysiloxane layer on the magnetite particles surface, can be seen from the diffraction patterns. Thus, an increase in polysiloxane shell for S1, SM1, SP1 samples may indicate the presence of broadened reflex at 10.5 (2-theta), the appearance of which can be attributed to the scattering of X-rays due to the presence of a certain amount of amorphous silica 36 . This reflex has low intensity with a minimum content of the polysiloxane layer ( Fig. 3 for S2, SM2, SP2 samples) in the diffraction patterns.
The crystal size of initial magnetite and nanocomposite particles was estimated from the XRD patterns by using Debye−Scherrer's equation 16 . The crystallite size of magnetite was found to be about ∼14.2 nm. The particle diameter is ∼16.4 nm for the samples S2, S3, SM2, SP2. However, the diameters of particles for S1, SM1 and SP1 samples calculated from this equation were 21.8 nm, 26.4 nm and 29.9 nm, respectively, indicating that a larger modifying layer was formed for the samples obtained with the catalyst. Fig. 4a confirmed that the pristine magnetite particles have a spherical shape and an average size of ∼12 nm (blue circle). The gray substrate which can be attributed to the polysiloxane part (green circle) of the composite 3,20 , can be observed. The same images can be noticed for similar magnetite/silica samples obtained by two and more stage, in which the magnetic particles are incorporated in polysiloxane shells 21 . From TEM photos the size of a magnetic particle with a shell was estimated, it was about ∼20 nm. The average particles size is in agreement with XRD data. Figure 4b shows the SEM microphotographs of the samples. The average diameters of spheres were calculated from these data presented in the Table 2. Based on TEM we found that the size of the functionalized particles is about 20 nm, however the spherical agglomerates are seen on SEM images. Largest agglomerates were obsereved for sample SM1. This may result from the higher condensation rate of MTES during synthesis. The particle size is gradually increased due to increasing size of a polysiloxane shell covering magnetic cores. Comparing these data with those obtained by photon cross-correlation spectroscopy (the last column in the Table 2), it can be concluded that for the first three samples are correlated. However, differences were observed for sample S1, where not only single magnetic particles, but also their agglomerates are incorporated into silica shell. Regarding to samples S2, S3, SM2 and SP2, the dimensions obtained from photon cross-correlation spectroscopy (PCCS) are almost   Low-temperature nitrogen adsorption/desorption isotherms. of the samples are shown in the  Table 2. It is noteworthy that the isotherms determined for each sample can be attributed to different types according IUPAC classification. Isotherms of bifunctional materials synthesized in the presence of catalyst (SM1 and SP1 samples), can be attributed to type IV; sample S1 -type I; the rest of the samples -type II. The sample S1 has the lowest surface area, which can be caused by the presence of agglomerates 360 nm and more in diameter. In the type IV of isotherm (bifunctional samples SM1 and SР1) a capillary-condensation hysteresis loop (Fig. 5) can be observed. This type of isotherms is typical for xerogels 37 , and indicates the presence of mesopores in the material structure. The way of adding a sol to the magnetite suspension, as well as the presence of Fcatalyst 29 during synthesis influenced the isotherm shapes and specific surface areas. Thus, for SM1 and SP1 samples, gradual introduction of sol to the mixture and the presence of NH 4 F, resulted in a rapid increase of the adsorbed N 2 at the initial region of isotherm higher values of the specific surface. This indicated the formation of a porous surface layer. The values of specific surface areas for S2, SM2, SP2 and S3 samples, does not significantly differ from the original magnetite 38 that can be attributed to a slight difference in the particle size of the samples.

TEM and SEM. TEM microphotographs for exemplary samples shown in the
Sorption study. The effect of initial pH on metal ions adsorption onto SP1 composite was studied for Pb (II) and Cd(II) solutions (Fig. FS2). The results demonstrate that the maximum of adsorption capacity appear at pH 5-6 for both metal ions. The highest sorption of lead ions at a similar pH is observed for mercapto functional groups 39 as well as for others 40 . pH was chosen ∼2 for mercury (II) and silver (I) ions adsorption on the basis of literature data for silica materials with thiol-containing functional groups 11 . The study of the adsorption kinetics of silver(I), mercury(II) and lead(II) ions for magnetically removable adsorbents allowed to determine the time needed to reach the equilibrium. The kinetic curves for the adsorption of metal ions are shown in the Fig. FS3. It can be seen that the equilibrium for Hg(II) and Ag(I) is reached in about 3 hours, and in about 6 hours for Pb(II). Thus, the above-mentioned times were chosen to carry out the adsorption measurements, leading to plot the adsorption isotherms.
It has been known that all thiols have an affinity for mercury ions as well as sulphides 25,41 . Therefore, it is not surprising that Me/Lig ratio is almost 1/1 for all studied samples. The constants calculated from the Langmuir and Freundlich equations, are presented in the Table 3. Regarding the calculated values, the equation of Langmuir isotherm is suitable only for SP1 sample. Ions' adsorption on the rest of materials surfaces is better described by the Freundlich equation. Interestingly, the shape of isotherms for the silver(I) ions are similar to the Langmuir type, thus they are better described by this equation for all samples. The adsorption capacities [in mmol/g] of Ag(I) ions are lowest, when compared with those obtained for Hg(II) ions. This may be due to the formation of disulfide groups at the surface of functionalized magnetite samples during storage, with which silver(I) ions cannot interact. This may be confirmed in the case of the sample S1 which has apparently the highest number of disulfide bridges and the lowest adsorption capacity, among S1, SM1 and SP1 samples. Furthermore, according to the analysis of the number of functional groups per 1 nm 2 , see Table 2, this sample had higher amount of functional groups onto its surface, when compared to other materials. Bifunctional materials, characterized by better developed surface, as well as the presence of the methyl or n-propyl groups, partially prevent the oxidation of 3-mercaptopropyl groups. Cadmium (II) ions can interact both 3-mercaptopropyl groups and disulfide bridges 42 , so Me/Lig ratio is close to 1/1. Presence of only part of available 3-mercaptopropyl groups were also confirmed by the results of adsorption of Pb(II) ions (Table 2 and Fig. 6). It can be seen that in this case Me/Lig are 1/2. Moreover, the obtained adsorption capacities of Pb(II) ions are lower that capacities obtained for Hg(II) and Ag(I) ions. In the case of the sample SP1 containing propyl groups, it may be caused by greater distance between particular mercapto groups present on silica surface.   Table 3. Parameters of mercury(II), silver(I) and lead(II) adsorption obtained by Langmuir and Freundlich isotherm equations. The distribution coefficients are presented in Table 3. Their values emphasize the high affinity of the surface layer of the bifunctional sorbents to the metal ions.
From the Table 4 it can be concluded that the synthesized samples possess high values of SC in relation of Cd(II), Hg(II), and Pb(II) ions, in comparison with the similar materials described in literature. Therefore, the methods of syntheses proposed in this paper make it possible to obtain magnetically removable adsorbents, characterized by high adsorption capacities in relation of heavy metals ions.
The selective adsorption of SP1 to capture other heavy metal ions in the presence of Ag + was studied using binary solutions containing equal concentrations of Ag + and Pb 2+ (or Cd 2+ ) at pH = 1.61 (Fig. 7a,b). The sorption was carried out from 1.2 M solutions due to this sample has 1.2 mmol/g thiol groups. We can summarize that this material showed very high affinity to Ag (I) in this condition. This is not surprising, since at such pH, as shown by our studies (Fig. FS2) and other authors 40 , the sorption of lead(II) and cadmium(II) ions is very low. Thus when all the reaction sites are occupied by silver(I) ions, the sorption of other metals does not occur. However, desorption and re-adsorption under such conditions decreases in the second cycle of using the adsorbent by about 30%, which is due to the poor solubility of silver sulphide.
We also observe that lead(II) adsorption is low in acid medium (about 12%), but about 44% of copper(II) is sorbed during sorption of divalent metals from mixtures with different pH and equal concentration (1.2 M). In   a neutral medium, the sorption capacity of both metal ions is the same (Fig. 7c). The sorption for these metals increases with increasing pH 14 . However, if sorption is carried out in acidic medium and the concentration of metals is reduced to microquantities, then silver is sorbed about 95%, and all other metals up to 20% (Fig. 7d). That is, based on the studies carried out, it can be concluded that the selectivity of metal ion extraction depends on the solution pH, ion concentration and the presence of free reaction sites.

Sorbent Cd(II) Hg(II) Pb(II) References
Wastewater treatment. Figure 8 shows the the removal efficiency of heavy metals from real wastewater in Ruskov, Slovakia. Accoding the law of Slovak republic for drink water parameters (N496 from 08.12.2010) concentration of Fe, As, Sb and Cr exceeds the permissible standards, the concentration of Cd is on the edge. Efficiancy of our materials we can estimate as we can see from Fig. 8, most metals are removed by 50%. The heavy metal ions with strong affinity to -SH such as Zn(II), Cd(II), and Pb(II), first bind with sorbtion sites, as the concentration of functional groups are large, there are many free sorption sites for other metals. Therefore, more 3-mercapto groups become accessible even to those ions with weaker affinity.
Adsorption of metal ions by recovered samples decreases, but especially for the monofunctional sample S1 (Fig. 8). This method of regeneration was selected in order to clean the surface of the sorbent as possible organic pollutants. The partial reduction of the sorbent can be due to formation of insoluble sulfur compounds, even in HNO 3 acid.

Conclusions
The magnetite nanoparticles coated with polysiloxane layers containing up to 2 mmol/g of mercapto groups were synthesized in a one-step synthesis method. It was revealed that the samples with bifunctional surface layer had twice more higher content of 3-mercaptopropyl groups than monofunctional produced in the same conditions due to the different rates of the silanes' hydrolysis. At the same time, alkyl groups can protect 3-mercaptopropyl groups from formation of disulfide bridges on the particle surface helping to maintain the functionality.
The effect of the catalyst presence in the magnetite suspension was studied. It was demonstrated that the stepwise adding of the sol and fluoride ions as catalyst contribute to increased number of thiol groups as well as better coverage of magnetite with polysiloxane layer characterized by well-developed surface. Such functional composite materials can be easily removed from the reactive medium using magnet wherethrough retain their magnetic properties. The synthesized functionalized magnetite nanoparticles can be used as effective solid phase extractants for Ag(I), Cd(II), Hg(II), and Pb (II) ions from solutions. Samples containing mercapto and alkyl groups together on the surface have shown higher removal efficiency of metal ions from real wastewater, such as up to 89% for Zinc(II), 90% for Cadmium(II), and 98.3% for Lead(II) ions. It was shown that after regeneration by nitric acid, the sample with 3-mercaptopropyl and propyl groups could be used for the second cycle of water treatment.
We began to develop methods of synthesis for samples S1, SM1, SP1 in 24 . The samples S2, S3, SM2, SP2 were prepared via similar procedure, but the distinctive feature being the addition of a sol to the magnetite suspension. This single-step procedure was carried out without catalyst.
Sample S1 (Fe 3 O 4 /SiO 2 /-(CH 2 ) 3 SH). The batch of Fe 3 O 4 (100 mg) was dispersed in 50 cm 3 of distilled water and treated by ultrasound for 10 min. The sol consisting of the silanes was prepared by prehydrolysis of TEOS and MPTMS (molar ratio 3/1). Then, the sol was added dropwise to magnetite suspension containing NH 4 F as catalyst (~0.5 сm 3 was added in every 15 min) during two hours, with continuous stirring with a mechanical stirrer. After addition of the last portion, stirring was continued for another 30 min. Therefore, the total stirring time was 2.5 hours. A dark-brown sediment was separated by magnet, rinsed three times with water (50 cm 3 ) and twice with ethanol (50 cm 3 ). The material was then dried at ~100 °C in an oven during 24 h.
Sample S2 (Fe 3 O 4 /SiO 2 /-(CH 2 ) 3 SH). Methods of preparation the magnetite suspension and functionalized sol were similar as for sample S1. Unlike to the previous synthesis, the sol of silanes was added to magnetite suspension dropwise during 20 min without catalyst. Consequently, the total mixing time was 6 hours.
Sample S3 (Fe 3 O 4 /SiO 2 /-(CH 2 ) 3 SH). Method of synthesis of this sample was the same as for the sample S2 with one exception -after prehydrolysis of TEOS, MPTMS was heated during 20 min, after that this sol was cooled and added to magnetite suspension dropwise.
Sample SM1 (Fe 3 O 4 /SiO 2 /-(CH 2 ) 3 SH/-CH 3 ) was synthesized according the method described for synthesis of the sample S1, except that 0.49 сm 3  For regeneration 10 cm 3 of 0.01 M HNO 3 solution containing 4% thiourea was added to the magnetic materials to desorb the silver.
Ag + , Pb 2+ , Cd 2+ , Zn 2+ and Cu 2+ concentrations before and after adsorption were determined using atomic absorption spectrometer С-115-М1 or with Varian AA 240 FS. The concentration of mercury(II) in the initial solutions and after adsorption process were determined by complexometric back-titration of the excessive EDTA with 0.025 М solution of MgSO 4 (eriochrome black T as indicator, ammonium buffer solution).
The metal ions sorption equilibrium data were correlated with both, Langmuir and Freundlich models. Langmuir isotherm is described by the following equation (1): where C eqv -the concentration of solute remaining in solution after equilibrium to be reached (mmol/L); a eqvthe amount of solute adsorbed in the same condition (mmol/g); a max -the maximum adsorption capacity in the monolayer and K L is the equilibrium constant of the adsorption process. The Freundlich isotherm is described by the following equation (2): where q -the amount of solute adsorbed (mmol/g); C eqv -the concentration of solute remaining in solution after equilibrium (mmol/L); K F -parameter related to maximum adsorption capacity in the multilayer of the adsorbent. Distribution coefficients (K d ) of the corresponding ions in case of microquantities sorption were calculated using the formula (3): where А o and А e are the initial and equilibrium concentrations of the ions in solution, respectively, mmol/dm 3 ; V is the liquid phase volume, cm 3 ; and m s is the sorbent weight, g.
Wastewater treatment. In order to evaluate the efficacy of these adsorbents, S1, SM1 and SP1 were tested in a real wastewater treatment. Wastewater first of all was filtered through a paper filter, and after that through 0.2 microm membrane filter. Initial pH was 8.26. An adsorption set of experiments was conducted the sample dose 2.0 g/dm 3 in 50 cm 3 capped polyethylene vials containing 25 cm 3 of effluent. The vials were placed inside constant orbital shaking at room temperature. The samples were collected 24 hours, filtered through paper filters, and the metal ions concentrations were quantified by Agilent Technologies 7700 Series ICP-MS. Nitric acid 1 M was used for the regeneration of the magnetic adsorbents (1 h). After that they were dried during 3 h at 100 °C and used again.
Characterization. The analysis for sulfur was performed at the KFK-2 equipment (Error ± 0.3 wt%).
Moreover, the estimation of the amounts of the ligand was carried out by a SEM-EDS technique using a Hitachi TM-1000-µ-DeX tabletop scanning electron microscope (Error ± 0.2 wt%). The analysis for carbon of bifunctional samples was performed by elementary analyzer Vario MACRO cube (Elementar Analysensysteme GmbH, Germany) using thermal conductivity detector (Error ± 0.2 wt%).
DRIFT spectra in the range of 4000-400 cm −1 were recorded on a spectrometer Thermo Nicolet Nexus FTIR using diffuse reflectance "SMART Collector" at a resolution of 8 cm −1 . Samples were mixed with KBr (for IR-spectroscopy, Aldrich) at 1:30 ratio. The data were processed by the software of spectrometer supplier company. A special thermovacuum Collector II attachment was used to register the IR spectra at higher temperatures.
The Raman spectra were recorded in the Raman microscope (inVia Reflex, Renishaw) with Raman dispersive system, using the 785 nm semiconducting laser. In order to avoid sample overheating 0.5 mW of laser power was used. The spectra of all samples were recorded at room temperature.
Thermal characterization of the prepared powders was studied using a Perkin-Elmer Pyris-1 TGA instrument connected with gas analysis unit operated with a Perkin-Elmer Spectrum 100 FTIR instrument (Error ± 0.02%).
The morphology of nanoparticles was characterized with the Analytical Scanning Electron Microscope JEOL JSM-6060 LA. In order to prevent the accumulation of surface charge and to increase the image contrast, the samples were covered with the shell of gold by cathode sputtering (JEOL Fine Coat, Ion Sputter JFC-1100). The nanoparticles were also characterized with the Transmission Electron Microscope JEOL: JEM-1230.
The granulometric analysis was carried out by using photon cross-correlation spectroscopy (PCCS) using a Nanophox particle size analyzer (Sympatec, Germany). A portion of aqueous suspension of each nanoparticles was diluted with the stabilizer to achieve a suitable concentration for measurement. The measurements were repeated three times for each sample.
X-ray powder diffraction (XRD) patterns were obtained using a Bruker SMART APEX-II diffractometer equipped with MoKα radiation (λ = 0.71073 Å). The diffraction was registered in rotation mode (1 degree/s) for powder samples sealed in 0.7 mm Lindeman tubes. Bruker APEX-II and EVA software were used for integration and data treatment. The low temperature nitrogen adsorption-desorption isotherms were recorded with a Kelvin-1042 instrument (Costech Microanalytical) at -196 °C. The samples were pre-degassed at 110 °C for 3 hours. The specific surface areas were calculated by the BET method.