INTRODUCTION

Although trace concentrations of certain heavy metals are metabolically needed, their unnecessary dosages may have serious side effects on critical organs such as brain, kidney, heart, liver, etc. [1, 2]. One of the trace metals in biological systems is cobalt. Apart from cyanocobalamin, the only known cobalt function, high levels of other cobalt resources are environmentally and biologically toxic [3, 4]. Cyanocobalamin is the metal component of vitamin B1. The genotoxic effect and oxidative DNA damaging of cobalt metal and its salts are distinguished. It is believed that Co(II) competes with Ca(II) for intracellular M(II) binding proteins and prevents Ca(II) entry. Moreover, cobalt can aggregate in the liver, kidney, pancreas, heart and reach a relatively high level in the skeleton and skeletal muscles [5]. Also, IARC (The International Agency for Research on Cancer) has classified cobalt as a possible human carcinogen [6].

As reported in the guidelines of WHO (World Health Organization) and EPA (Environmental Protection agency), among all employed methods for removing heavy metal ions, solid-phase extraction (SPE) has been preferred for treating water and wastewater from the environmental and economic points of view [7-12]. Simplicity, flexibility, low usage of organic solvents, low cost, and short extraction time are some of the considerable features of SPE, which make this technique superior to other conventional methods. Furthermore, adsorbents can be regenerated by desorption processes and reused [13].

It should be noticed that the success of SPE relies on the choice of an appropriate adsorbent. In this respect, MNPs have been engaged for supporting functionalities with coordination ability to toxic metal ions [14]. During the last decades, magnetite nanoparticles (MNPs) have attracted much attention in the catalysis area owing to their stability, low toxicity and cost, high dispersibility, good biocompatibility, and high surface-to-volume ratio [15-19].

Among the wide range of chemical procedures that deal with the extraction of metal ions from aqueous media, surface-functionalized MNPs with metal coordinating ligands have attracted increasing interest. For example, Lin et al. [8] prepared magnetic polymer beads containing amino groups on their surfaces in order to enable them to capture Cu(II), Ni(II), and Co(II) cations from aqueous solutions. They observed the highest adsorption capacities over the pH range of 3 to 5, with slight temperature dependence. The clearance and regeneration efficiencies of their magnetic adsorbent were reported to be higher than 98%. Separation of rare-earth ions by surface-treated MNPs, SiO₂, and TiO₂ NPs with N-[(3-trimethoxysilyl)propyl]ethylenediamine triacetic acid (TMS-EDTA) has been reported by Dupont et al. [20]. Enhanced metal uptake capacities (100 to 400 mg g^-1) of the investigated NPs are attributed to the small diameter and high surface area. They claimed that silica and titania NPs could be loaded by higher amounts of TMS-EDTA and, therefore, larger quantities of rare-earth ions can be adsorbed, but their removal from solutions is very tedious. Jainae et al. [21] prepared thiol-functionalized polystyrene-coated magnetic CoFe₂O₄ NPs for selective adsorption of Hg(II) cations. By employing a concentrated Hg(II) solution (100 ppm), to ensure full coverage of Hg(II) ions, they reported the metal uptake capacity of 89.6 mg g^-1 and ten cycles of usage. Qiu et al. [22] reported simultaneous detection and removal of low Cu(II) concentrations by a magnetic nanocomposite modified by a selective chelating group with the removal efficiency of 80% and 27.5 mg g^-1 adsorption capacity. Recently, Chen et al. reported removal of Co(II) and Ni(II) aqueous solution using reductive self-assembly of three-dimensional magnetic fungal hyphal/graphene oxide nanofibers [23]. A continuous-flow recycle reactor was also proposed for emergency aqueous solution treatment. Wang et al. used carboxymethylcellulose-chitosan film modified magnetic alkaline Ca-bentonite for the efficient removal of Pb(II) and Cd(II) from aqueous solution. The sorbent could be efficiently recycled with less decrease in adsorption capacity [24].

In this study, a birhodanine containing silane compound, i.e. [(E)-5-(3-(3-(trimethoxysilyl)propyl)-4-oxo-2-thioxothiazolidin-5-ylidene)-3-phenethyl-2-thioxothiazolidin-4-one] (TMOS-BIRD), is studied to modify the adsorbent’s surface. Diverse biological activities through inhibition of numerous targets, e.g., HIV-1, Alzheimer’s disease, HCV NS3 protease, b-lactamase, PMT1 mannosyl transferase, PRL-3, and JSP-1 phosphatases are reported for rhodanine-based molecules [25, 26]. Rhodanine heterocycle can also coordinate with metal ions due to the presence of thioamide and keto groups. In a study, Song et al. [27] grafted polyrhodanine onto the inner surface of an anodic alumina membrane to remve mercury, lead, and silver ions. They attributed the maximum removal capacity (7.01 mmol/g polymer) to Ag(I) cations. Lee et al. [28] prepared rhodanine-loaded poly(methylmethacrylate) electrospun nanofibrous membranes to remove heavy metal ions. Their experimental results suggested that their prepared membranes exhibit good ion uptake capabilities for Ag(I) and Pb(II) cations. Amiri et al. [29] proposed chitosan/polyrhodanine nanocomposite particles to eliminate Ni(II) ions from aqueous solutions. Their observed adsorption trend followed pseudo-second-order kinetics and gave the maximum equilibrium uptakes of 61.43 and 67.70 mg g^-1, obtained from the Langmuir and Sips isotherms, respectively. Chen et al. [30] used self-assembly of rhodanine/gold NPs to construct a colorimetric sensing device for Hg(II) detection in aqueous media. The aggregation of NPs is induced by Hg(II) ions, and a color change from wine red to blue is observed. Based on the experimental results, particle aggregation is due to the formation of the [Hg(rhodanine)₂]²⁺ adduct. 

We reported the preparation of a rhodanine-based monomer. The monomer was polymerized on MNPs via the SI-ATRP approach [31]. The polymer grafted NPs were used to quickly eliminate Co(II) ions, and 86.0% of Co(II) ions were removed in only 7 min.

Birhodanine compounds with the exact IUPAC name of 2,2’-dithioxo-[5,5’]bithiazolidinylidene-4,4’-dione) are an interesting group of rhodanine-based derivatives and can be synthesized through various routes [32-34]. The use of these organic heterocycles has been extended to numerous applications, such as dyes [35], transistors [36], catalysts [37], polymers, and nanocomposites [38]. In this work, we wish to report the design and synthesis of a new birhodanine compound for surface silylation. The prepared compound has two structural rhodanine heterocycles. It can merge enhanced coordinating activity with easy surface anchoring procedure on metal oxides to produce novel heterogeneous adsorbents for removal of metal ions. It must be noted, ease of hydrolytic cleavage of rhodanine nucleus in alkaline media limits the incorporation of the supported catalyst to neural and weak acidic solutions [38]. This can be considered as the main disadvantage of the supported adsorbent. To the best of our knowledge, the present study is the first report on the immobilization of rhodanine moiety via silylation. 

Following our approach reported for one-pot synthesis of birhodanine compounds under solvent-free condition [39, 40], and considering the rhodanine ring’s ability towards coordinating to metal ions, we thought that anchoring this structure on the surface of MNPs might be invaluable for the removal of metal ions. Therefore, in this work, we report the preparation and characterization of a new birhodanine containing silylating compound (TMOS-BIRD; Fig. 1) that can be incorporate for surface silylation of metal oxides such as MNPs.

The adsorption performance of the prepared MNPs was assessed via removing Co(II) cations from aqueous solutions. Also, the kinetics and thermodynamics of adsorption are investigated under optimal conditions. Furthermore, the Langmuir and Freundlich isotherms are employed to fit the experimental adsorption data.

MATERIALS AND METHODS

Synthesis of the silylating birhodanine compound (TMOS-BIRD)

A mixture of APTMS (0.358 g, 1 mmol), carbon disulfide (0.304 g, 4 mmol), compound VI (0.616 g, 2 mmol) and a catalytic amount of tetrabutylammonium bromide (TBAB) (0.128 g, 0.4 mmol) was stirred for 1 h at room temperature, under solvent-free condition. Then, water (8 ml) and dichloromethane (32 ml) were added. The organic layer was evaporated after drying over anhydrous calcium chloride, and the crude product was recrystallized from EtOH to afford 0.316 g golden crystals (Yield: 30%) [40]. m.p. 205-207°C. FTIR (KBr, cm^-1): 2927 (m), 1695 (s), 1428 (m), 1349 (s), 1176 (s), 1086 (s), 870 (m), 743 (m), 699 (m). ¹H-NMR (250 MHz, CDCl₃) δ: 7.26-7.31 (m, J =5 Hz, 5H), 4.33 (t, J =7.5 Hz, 2H), 4.11 (t, J =7.5 Hz, 2H), 3.56 (s, 9H), 2.99 (t, J =7.8 Hz, 2H), 1.70-2.00 (m, 2H), 0.67 (t, J =7.5 Hz, 2H) ppm. ¹³C NMR (62.5 MHz, CDCl₃) δ: 194.2, 166.6, 136.8, 128.7, 127.0, 124.9, 124.3, 50.6, 46.7, 45.6, 33.0, 20.5, 13.7, 6.5 ppm. 

Silylation of magnetite and synthesis of MNP@BIRD NPs

TMOS-BIRD (0.700 g, 1.3 mmol) was added to a dispersion of MNPs (0.600 g in 30 ml) in 6.0 ml of 2M triethylamine solution in toluene and stirred for 24 h under argon at room temperature. The silylated MNPs were decanted magnetically and washed thoroughly with MeOH. Removal of any un-reacted TMOS-BIRD was ensured by repeating the washing procedure three times, and finally, the silylated NPs were dried in a vacuum [41, 42].

Adsorption study

Adsorption investigations were accomplished by the batch technique in the dispersion of MNP@BIRD NPs in aqueous solutions of Co(II) ions. NaOH and HCl (0.1 M) were used to adjust the pH of aqueous solutions. The details of adsorption study, kinetic study, equilibrium isotherms, thermodynamic assessment, and desorption experiments were explained in the supporting information file. 

RESULTS AND DISCUSSION

Synthesis and characterization of MNP@BIRD NPs

The silylating birhodanine compound, i.e., TMOS-BIRD or compound (VI), was prepared according to Fig. 2, and it was used for the silylation of MNPs. Methyl 2-(4-oxo-3-phenethyl-2-thioxothiazolidin-5-ylidene)acetate (IV) was easily obtained (55 % yield) through the one-pot three-component reaction of 4-phenethylamine (II), carbon disulfide (I), and DMAD (III) in water at room temperature [43]. Compound IV was further reacted with carbon disulfide (I) and APTMS (V) in the presence of TBAB at room temperature under solvent-free conditions to give TMOS-BIRD (VI).

The ¹H NMR and ¹³C NMR spectra of TMOS-BIRD are exhibited in Figs. 3a and 3b, respectively. The aromatic ring protons of TMOS-BIRD appeared at 7.20 to 7.40 ppm, and the protons related to the methylene groups are distinguished with the expected multiplicities, Fig. 3a. In Fig. 3b, 16 signals are observed that are consistent with the structure of TMOS-BIRD. In the mass spectrum (Fig. 4), molecular ion peak is appeared at m/e = 528.2 (4%).

As aforementioned above, MNP@BIRD NPs were obtained through silylation of MNPs by TMOS-BIRD. The FTIR spectra of MNPs, MNP@BIRD NPs, and TMOS-BIRD are shown in Figs. 5a to 5c, respectively. In the FTIR spectrum of MNP@BIRD NPs (Fig. 6b), the appearance of the 3030 cm^-1 (ν_s arom. C-H), 2928 cm^-1 (ν_s CH₂), 1696 cm^-1 (C=O stretching), and 1428 cm^-1 (CH₂ bending) transmittance bands, along with the rhodanine ring related vibrations at 1350, 1262 and 1176 cm^-1, clearly confirms silylation of MNPs by TMOS-BIRD. The presence of the siloxane vibrational mode (ν_as Si-O-Si) at 1112 cm^-1 beside the characteristic peak of magnetite at 588 cm^-1 (44) confirms the success of the silylation reaction and formation of MNP@BIRD core-shell. 

In the SEM image of MNP@BIRD NPs, the spherical morphology of 20-30 nm diameter can be distinguished, Fig. 6. The TEM image of MNP@BIRD reveals spherical or ellipsoidal-shaped NPs of uniform size distribution, Fig. 7. Therefore, the prepared MNP@BIRD NPs are superparamagnetic, and their size distribution is equal to the recommended critical value (25 nm) for biomedical applications [44, 45]. As seen in Fig. 7, MNPs are covered by a silica layer of about 10.1 nm diameter. Twenty selected random particles in a TEM Image were used to determine the silica layer thickness by ImageJ software.

TGA was used to estimation of the loaded amount of the sorbent. Water removal is shown by a mass loss of about 2.0% for the bare MNPs from ambient temperature to 600ºC, Fig. 8a. TGA curve of MNP@BIRD exhibits a mass loss of about 6.8% under a similar condition due to thermolysis of the birhodanine moiety, Fig. 8b. The difference between these two values (Δm = 4.8%) can be attributed to the amount of grafted birhodanine molecules onto MNPs (4.8% or 48 mg g^-1 of MNPs).

The XRD patterns of the bare MNPs and core-shell of MNP@BIRD are presented in Figs. 9a and 10b. Both samples exhibit the characteristic peaks of cubic Fe₃O₄ with the Fd-3m space group [46]. Therefore, the spinel lattice structure of magnetite is retained. The reduction of peak intensities can be attributed to the coating of MNPs by an amorphous silica layer, Fig. 9b.

The hysteresis curves of MNPs and MNP@BIRD NPs were depicted in Fig. 10. The saturation magnetization (M_s) of MNPs and MNP@BIRD NPs are 69.5 and 78.1 emu/g, respectively. The obtained values are less than the value reported for bulk Fe₃O₄ particles [47]. Therefore, the prepared MNPs are superparamagnetic, and covering MNPs by a non-magnetic silica layer is responsible for decreasing of M_s value for MNP@BIRD NPs. Moreover, the small field coercivity of MNP@BIRD NPs (H_c = 36.5 Oe) shows that the prepared NPs are superparamagnetic [48]. Therefore, the magnetization of MNPs did not affect significantly by the silylation reaction. The results for magnetization measurements are summarized in Table 1.

Adsorption Study

Rhodanine compounds contain several potential coordinating sites to attach to different metallic ions [49, 50]. Therefore, immobilization of rhodanine derivatives onto magnetic substrates for removing toxic heavy metal ions from water or wastewater is worthful. To demonstrate the efficiency of the prepared adsorbent for Co(II) removal, MNP@BIRD NPs were added to the solution of Co(II) ions at room temperature and then collected by an external magnet. Figs. 11a and 11b show the FTIR spectra of MNP@BIRD NPs and the Co(II) complexed adsorbent, respectively. The frequency related to C=O stretching in the Co(II) complexed adsorbent shows a very distinct shift (by 30 cm^-1) and appeared at 1638 cm^-1. The observed red shift is due to the weakening of the C=O bond upon coordination, which lowers the stretching frequency. The possible coordination of Co(II) ions to the supported birhodanine moieties is demonstrated in Fig. 12.

Metal removal can also be observed visually by receding the purple colour of the solution after magnetic decantation, Figs. 13a-c. The optimal adsorption conditions were determined in order to investigate the corresponding adsorption kinetics and thermodynamics. Contact time of 30 min, pH 8, initial metal ion concentration of 3.0 ppm, and 10 mg of the adsorbent were chosen as the best conditions for Co(II) removal from a 25 ml aqueous solution. The details of the optimization trials were described in the Supporting Information file.

The Study of Adsorption Kinetics 

Figs. 14a and 14b show the experimental results of the kinetic investigation. Lagergren’s pseudo-first-order (Eq. 1) and Ho’s pseudo-second-order (Eq. 2) models were applied to fit the adsorption kinetic data [41]. 

where k_f (1/min) and k_s (g/(mg. min)) are the pseudo-first-order and pseudo-second-order equilibrium rate constants, respectively. Q_e and Q_t are adsorbed metal ions (mg g^-1) at equilibrium and time t, respectively. For the pseudo-first-order model, Q_e, k_f, and correlation coefficient obtain from the linear plot of ln (Q_e - Q_t) against t, Fig. 14a. The correspondent values for pseudo-second-order equilibrium (Q_e, k_s, and R²) obtain from the plot of t/Q_t versus t, Fig. 14b. The results are summarized in Table 2. As can be seen, the pseudo-first-order model is better fitted with the experimental data than that of the pseudo-second-order model. Furthermore, the calculated and experimental values of adsorption capacity (Q_e,cal and Q_e,exp) are more consistent in the case of the pseudo-first-order model (Q_e,exp = 5.36 mg g^-1, Q_e,cal = 6.71 mg g^-1).

Adsorption isotherm studies

The adsorption parameters were found by fitting the experimental results with the Langmuir and Freundlich isotherm models. The isotherm model that best fitted the experimental data was then used to calculate the adsorption parameters. According to the Langmuir model, the binding sites of the adsorbent are identical. The adsorbed molecules are single-site located, and they have not any interaction with other molecules. Therefore, adsorbate species form a monolayer on the surface. The linear form of Langmuir isotherm can be written as Eq. 3:

where C_e, Q_e, Q_m, and K_L are Co(II) equilibrium concentration (mg/L), adsorbed metal ion at equilibrium (mg g^-1), the maximum capacity for Co(II) uptake (mg g^-1), and the Langmuir adsorption constant (L/mg), respectively. Q_m and K_L are the Langmuir model parameters and obtain from the linear plot of C_e/Q_e against C_e. 

The linear form of Freundlich isotherm is written as Eq. 4:

where Q_e is the adsorbed Co(II) ions onto the sorbent at equilibrium (mg g^-1), C_e is the Co(II) concentration at equilibrium (mg/L), and K_F is the Freundlich isotherm constant
((mg g^-1)(L/mg)^1/n). The values of K_F and 1/n (sorption intensity) obtain from the linear plot of ln Q_e vs. ln C_e [51]. 

The plots for Langmuir and Freundlich isotherms are shown in Figs. 15a and 15b, respectively. The related parameters are also listed in Table 3. As can be seen, Langmuir isotherm shows better compatibility with the experimental data (R² = 0.9964) than the Freundlich isotherm (R² = 0.9072). This implies that a layer of the metal ions covers the MNP@BIRD surface. The maximum adsorption capacities, obtained from the Langmuir (Q_m = 6.02 mg g^-1) and Freundlich (K_F = 5.31 mg g^-1) models, are both following the experimental value (Q_e.exp = 5.36 mg g^-1). Fig. 16 presents the calculated R_L values at different initial Co(II) concentrations, and the results are summarized in Table 3. As seen, the R_L values are in the range of 0 and 1. This means that the adsorption Co(II) ions on MNP@BIRD NPs is a favorable process [52].

Adsorption thermodynamics

Thermodynamic parameters are essential for a practical application because they provide useful information about the nature of surfaces and adsorbed phases. In general, ΔH and ΔS values can be calculated from Eqs. 5 and 6:

where K_C , R, and T are the equilibrium constant, the gas constant (8.314 J/(mol.K)), and absolute temperature (K), respectively. The values of ΔH and ΔS were derived from the linear plot of ln K_C against 1/T, which is presented in Fig. 17.

ΔG can be obtained from Eq. 7.

The results of the thermodynamic investigation are summarized in Table 4. Positivity of the ΔH value indicates the endothermic nature of Co(II) adsorption onto MNP@BIRD NPs, whereas the positive ΔS value declares the increase of randomness at the solid/liquid interface during SPE of Co(II) ions. Releasing of water molecules around the hydration sphere of Co(II) ions upon adsorption of the solvated metal ions is responsible for the positive ΔS value. Meantime, the negative ΔG values reveal that the adsorption process is spontaneous. Therefore, ΔG is mainly influenced by TΔS, and increasing entropy favors the adsorbent interaction with the metal ions [10].

Recyclability of the sorbent

Preparing a synthetic adsorbent is usually done via a complicated process. Therefore, sufficient attention to the recycling of a used sorbent is necessary. In the present study, regeneration of the used MNP@BIRD NPs was done by treating the used adsorbent with the diluted aqueous HCl solution followed by neutralization and washing steps. As shown in Fig. 18, the values of extraction efficiency (E %) and recovery % were not changed significantly after seven consecutive runs. The average extraction efficiency (E %) reached to 62.0% (89.1 % recovery) after seven cycles. 

The adsorption capacity of MNP@BIRD NPs is compared with a few literature-reported adsorbents in Table 5. Comparison of the outlined adsorption capacities demonstrates that in the most cases, the prepared NPs are more effective than the reported adsorbent (Q_m = 6.02 mg g^-1). The obtained NPs also enjoy magnetic decantation as the main advantage.

CONCLUSION

The synthesis and characterization TMOS-BIRD as a new silylating agent for the silylation of MNPs have been reported. TGA experiments reveal that the birhodanine content of the prepared NPs was 48 mg g^-1. The prepared MNP@BIRD NPs were used for magnetic deionization of the aqueous solutions from Co(II) ions. The calculated metal adsorption capacity was 6.02 mg g^-1, and the experimental data obeys the pseudo-first-order kinetics and Langmuir isotherm model. Thermodynamic investigation revealed the endothermic and spontaneous nature of the adsorption process. The recyclability of the prepared sorbent was shown at least seven times. Based on the obtained results, it can be suggested that MNP@BIRD NPs are stable, reusable, and can be efficiently used in water and wastewater treatment processes. 

ACKNOWLEDGEMENT

The Graduate Council of the University of Mohaghegh Ardabili is gratefully acknowledged for providing the financial support. This work was supported by University of Mohaghegh Ardabili (Grant number 222/13D/94). 

CONFLICT OF INTEREST

The authors declare that there is no conflict of interests regarding the publication of this manuscript.