Adsorption of Transition Metal Chlorides by Silica with Grafted 1n-Propyl-3-Methylimidazolium Chloride

Silica with grafted 1-n-propyl-3-methylimidazolium chloride was investigated as adsorbent for the removal of cobalt(II), copper(II), zinc(II), cadmium(II) and mercury(II) chlorides from alcoholic and aqueous solutions. The adsorption proceeds via the formation of metal-chloride anionic complexes detained near the grafted organic cations as counterions. This conclusion is corroborated by the electronic and Raman spectra. The high removal degrees of the metal chlorides from solutions are easily reached. The adequate models of adsorption equilibria consider dissociation and auto-complexation of chlorides in solutions and formation of triand tetrachlorometallate anions in the solid phase. Also, the equilibrium constants for the CoCl2 and CuCl2 adsorption by several materials containing grafted organic cations and chloride counterions were recalculated in the framework of this model. The weak dependence of equilibrium constants on the nature of immobilized cations and carriers agrees with the established chemistry of metal chlorides fixation.


Introduction
Silica-organic hybrid materials containing complexing organic species are widely used for preconcentration and further determination of various pollutants, creation of catalysts, electrochemical and optical sensors.To modify purposefully the properties of materials, one needs information about the stoichiometry of interactions of organic modifiers with objective sorbates as well as the thermodynamic stability of the formed complexes.This information is also important for revealing useful features of hybrid materials and forecasting their properties at the stage of synthesis.
][3][4][5][6][7][8][9] They demonstrate ion exchange capability with respect to different anions. 6A noteworthy feature of these materials is their ability to adsorb metal halides.The equilibria of metal chlorides adsorption from solutions have been studied by means of quantitative physicochemical analysis. 10The obtained information about composition and thermodynamic stability of surface complexes has allowed to solve the practically important task of metal chlorides extraction and preconcentration from ethanol solutions.The formation of the metal-chloride anionic complexes has been shown to be the key aspect of this extraction. 6,11ecently a new hybrid material, silica with grafted ionic liquid 1-n-propyl-3-methylimidazolium chloride, has been obtained and used for removal of anionic dyes from aqueous solutions. 12,13Likewise other materials with fixed heterocyclic cations and chloride counterions, this hybrid material is also promising as a sorbent for metal chlorides removal from solutions.5][16] This means the possibility to prepare heterogeneous catalysts based on silica modified with 1-n-propyl-3-methylimidazolium chloride through the sorption of metal chlorides.With this Vol.26, No. 6, 2015   in mind, the comprehensive information on equilibria of metal chlorides adsorption becomes necessary for adjusting the composition and surface concentration of fixed metalchloride anionic complexes.
In this paper we discuss the adsorption of cobalt(II), copper(II), zinc(II), cadmium(II) and mercury(II) chlorides from methanol, ethanol and aqueous solutions by the silica chemically modified with 1-n-propyl-3-methylimidazolium chloride.As distinct from the previous studies, special attention is paid to complexation and auto-dissociation equilibria of metal chlorides in solutions.This makes it possible to present a more realistic chemistry of metal chlorides binding and to determine the corresponding equilibrium constants.The composition of fixed metalchloride complexes concluded from the simulation of equilibria is corroborated by the results of visible and Raman spectroscopy.Also, the models of CoCl 2 and CuCl 2 adsorption by earlier studied hybrid materials have been refined.

Material under study and plausible chemistry of metal chlorides binding
Silica with grafted 1-n-propyl-3-methylimidazolium cations (denoted below as Cat + ) and bearing exchangeable chloride counterions (the SMICl material, Figure 1) has been prepared through chemical modification of silica surface by 3-n-propyltrimethoxysilane-1-methylimidazolium chloride.The synthetic and characterization procedures have been described elsewhere. 13e specific surface area of the material was 220 ± 5 m 2 g -1 , and the specific concentration of grafted 1-n-propyl-3-methylimidazolium chloride was 1.00 ± 0.02 mmol g -1 .The average pore diameter and the specific pore volume were 12.4 nm and 0.88 cm 3 g -1 , correspondingly.The aqueous suspension of material (1 g L −1 ) has pH 5.0.
By its chemical nature, the SMICl material belongs to the ion exchangers (it contains exchangeable chloride ions).However, the consequent chemical mechanism of metal chlorides fixation by this material does not agree with the formal ion exchange models.It was shown 6,11 that binding of metal chlorides from ethanol solutions occurs due to the formation of anionic metal-chloride complexes.In the case of metal chlorides under study, the formation of anionic complexes MCl 3 -and MCl 4 2-is possible (M is Co, Cu, Zn, Cd or Hg).So, the overall process of the metal chlorides binding assumes the transfer of metal species from the solution to the solid phase and includes interactions of two types: the formation of metal-chloride anionic complexes (formation of covalent bonds) as well as the purely electrostatic retention of the formed anions near the grafted 1-n-propyl-3-methylimidazolium cations.The general case may be presented by the following equations: The SMICl material with the sorbed metal chlorides was characterized by means of visible diffuse reflectance and Raman spectroscopy.The diffuse reflectance spectra have been recorded with the Varian Cary 500 ultravioletvisible (UV-Vis) near infrared (NIR) spectrophotometer (Agilent Technologies); spectra of the material with copper(II) sorbed from aqueous solutions were recorded with the Hitachi 850 spectrofluorimeter equipped with an add-on device for measuring reflection.BaSO 4 was used everywhere as the standard of reflection.The values of the Kubelka-Munk function (F) were calculated from the measured reflections (0 < R < 1) as The Raman spectra were registered with Horiba TXplora, source of 514 nm and 16.4 mW cm -2 power.
No sorption of metal chlorides by unmodified silica was observed.The batch technique was used to measure the adsorption of metal chlorides by the SMICl material at 298 ± 2 K.To study the sorption of metal chlorides from solutions, the precise weights (ca.50 mg) of the SMICl material were immersed into 50.0mL of alcohol solutions or 30.0 mL of aqueous solutions with various concentrations of metal chlorides.The mixtures were shaken for 3 h at 298 K and stayed in closed flasks for 2 days.The residual amounts of metal species (except Hg) in solutions were determined by titration with the ethylenediaminetetraacetic acid (EDTA) according to the procedures described elsewhere. 17The residual concentrations of mercury(II) in solutions were determined spectrophotometrically with diphenylcarbazone. 18Absorbances were measured at 520 nm with the use of photoelectric colorimeter KFK-3 (Zagorsk optical and mechanical plant, Russia).The experiments were repeated at least triply.The relative standard deviations of the measured metal concentrations in solutions were in the range 0.5-3%.
The values of metal species adsorption (N f , mol g −1 ) were determined from the differences between their initial and equilibrium concentrations in solutions as where t(M(II)) and [M(II)] are the initial and equilibrium metal species concentrations (mol L −1 ), correspondingly; V are the solution volumes (L); and m s are the masses of the material (g).
As distinct from another studied systems, the adsorption of CuCl 2 from aqueous solutions proceeds reluctantly.This is explained by the weak stability of the copper(II)chloride complexes in aqueous media.As a result, equilibria of the copper(II) adsorption are shifted to the left side.According to the Le Chatelier's principle, the increase of the Cl − concentration will favor the copper(II) adsorption.Thus, to enhance the copper(II) binding, the great excess of KCl (0.14-0.16 mol L -1 ) was created in solutions.Under such conditions the formation of fixed copper-chloride anionic complexes becomes noticeable by the appearance of the yellowish-brownish coloration of the solid.At the same time, the values of the copper(II) adsorption remain low.They are found according to equation 3 from rather small differences between the initial and equilibrium concentrations of copper(II) in solutions.To obtain reasonably exact N f values, one should measure these concentrations with the improved accuracy.This was achieved by the replacement of the B class glassware with the more precise A class and the repetition of the experiments at least 9 times.As a result, the relative standard deviations of the copper(II) concentrations were reduced to 0.2%.The primary experimental data (the dependencies of the equilibrium concentrations [M(II)] on initial amounts of metal chlorides in solutions) as well as their fits by the constructed models of adsorption equilibria are presented in the Supplementary Information (Figures S1-S5).

Simulation of adsorption equilibria. Calculation of fitting parameters
In the previous works the equilibria of the MCl 2 adsorption have been simulated under the assumption that metal chlorides in ethanol solutions 1,2,4,5,7-9 and HgCl 2 in aqueous media 3 are not dissociated.On the contrary, the complete dissociation of ZnCl 2 and CdCl 2 in aqueous solutions was assumed. 3A more rigorous approach requires the account of states of metal species in solutions (dissociation and autocomplexation of metal chlorides).Also, it is necessary to keep in mind that different metal containing anions may be present at the surface simultaneously.
Owing to the mentioned features of the systems under study, the stoichiometry of interactions responsible for the metal chlorides binding may be rather complicated.In addition, cooperativity and energetic heterogeneity effects may influence the adsorption process.Thus, the conventional Langmuir adsorption model is applicable only as an exception, namely, when the undissociated metal chlorides predominate in solutions and metal-chloride anionic complexes of only one type are formed on the surface.Under such circumstances equations 1 and 2 are simplified as follows: and the equation of adsorption isotherm takes the form where t Q is the effective sorption capacity (mol g -1 ), [MCl 2 ] is the equilibrium concentration of MCl 2 in solution (mol L -1 ) and K L are the equilibrium constants.
To simulate more complicated equilibria, more sophisticated models are required.In this work, the preference was given to the model of polydentate binding. 4,10This model is a special case of a general lattice model which represents the surface of the material as an assemblage of binding centers (∼ Cat + Cl -) Z .Each center contains Z active groups ∼ Cat + Cl -and is able to bind from 1 to Z sorbate species.This process is described by Z equilibrium constants σ i (Z) , where i is the number of fixed sorbates.So, the fitting parameters of the model are the size of binding centers Z, their specific concentration T (mol g -1 ) and the equilibrium constants σ i (Z) , i = 1, 2, ..., Z. Figure 2 represents one of the variants of this model for Z = 2.
The construction of the model starts from the lowest possible Z value.The respective values of equilibrium constants are calculated through the minimization of a fitting quality criterion (equation 7), and the statistical adequacy of the model is checked in a routine way (e.g., using χ 2 criterion).If the experimental data are not described within their errors, Z is increased and a new calculation of σ i (Z) is performed.This procedure is repeated until the adequate description is achieved.
The fitting parameters are determined through the approximation of the composition-property dependencies.The criterion to be minimized is where N is the total number of experimental points, weighted discrepancy , A is the value of the measured property of the equilibrium system, the statistical weights of measurements w k = 1/σ k 2 , and σ k 2 is the variance of A k experimental evaluated from the model of errors of experimental data.The simplest reasonable model of errors capable to equalize the influence of all measurements on the estimations of fitting parameters presumes that values of fitted property A are measured with the same relative standard deviation σ r .For example, if equilibrium metal concentrations in solutions [M(II)] are the measured properties, the weights are assigned as For a certain fixed T value, the estimations of σ i (Z) corresponding to the minimum of criterion in equation 7 are calculated iteratively.In this work all simulations have been performed with the aid of our software program CLINP 2.1. 19Simultaneously with the σ i (Z) estimates, their dispersion-covariance matrix is obtained.The model is considered as adequate if inequality holds, where χ 2 f (5%) is the 5% point of the chi-square distribution with f = N -Z degrees of freedom.
As a rule, several models with fixed Z value rather than the unique model of Z-dentate binding fit the compositionproperty dependence within the limits of experimental errors.The adequate models differ in the accepted values of the specific concentration of binding centers (any T value within a certain interval provides adequate fitting).To find the most reliable averaged T estimate ( -T ), we have applied the recently described procedure, 20 that gives proper weights to all constructed adequate models.The σ i (Z) values corresponding to the -T estimates are reported throughout the paper.

Results and Discussion
To understand better the chemistry of metal species fixation, it is useful to consider their state in solutions.If the initial concentrations of metal and chloride ions as well as the stability constants of metal-chloride complexes are known, the equilibrium composition of solutions is easily calculated.We have been performed such computations using the stability constants listed in Table 1.
Prior to discuss the results in total, the simulation details will be explained for the case of CuCl 2 adsorption from ethanol solutions.
The calculated degrees of formation of copper(II) species in solutions are shown in Figure 3. Dependently on the CuCl 2 concentration, solvated Cu 2+ cations, complexes CuCl + or CuCl 2 predominate in solutions (i.e., the supposition about the small degree of the CuCl 2 dissociation is too rough), while the higher anionic chloro complexes are the minor species at all studied CuCl 2 concentrations.
According to the results of the previous studies 6,11 and the spectral data (Figure 6), the presence of copper(II) at the surface in the form of anionic complexes CuCl 3 -and CuCl 4 2was supposed.Correspondingly, the simplest version of the model of polydentate binding presumes Z = 2. Thus the reactions responsible for the copper(II) fixation may be written as follows:  the interval 0.17-0.45mmol g -1 .The averaged estimate -T = 0.28 mmol g -1 , and the corresponding equilibrium constants are as follows: log σ 1 (2) = 9.75 (0.04), log σ 2 (2) = 18.32 (0.16) (here and below the values in the parentheses are the standard deviations).
Thus constructed models of the metal chlorides binding for all studied systems are presented in Table 2.The models fit adequately the primary experimental data (Table 2 and Figures S1-S5 in the Supplementary Information).The models differ from the conventional Langmuir adsorption model by the explicit account of different coexisting complex species in the solution and on the surface.Just as the Langmuir model, they do not take into account a number of complicating effects that often affect the adsorption processes, such as multilayer adsorption, energetic heterogeneity of binding centers or deviation from the strict stoichiometry of interactions.The adequacy of the found models points to the negligibility of all mentioned effects.
Based on the found sorption capacities and equilibrium constants, we have calculated the equilibrium compositions of all studied adsorption systems and used these data to find the degrees of formation of different species on the surface at different concentrations of metal species in solutions.The degrees of formation of metal containing species (Cat + Cl -) Z M j Cl k (total charges are not shown) in the adsorption systems were calculated as where t(M(II)) is the initial MCl 2 concentration (mol L -1 ); j and k are the stoichiometric indices (for instance, CuCl + in solution is presented as (Cat + Cl -) 0 Cu 1 Cl 1 while the stoichiometry of the surface complex containing anion CuCl 4 2-is written as (Cat + Cl -) 2 CuCl 2 ); the square brackets denote the equilibrium concentrations (molar for species in solutions and specific for species on the surface); for species in the solution phase F i = 1, Z = 0; for fixed species F i = m s / V, Z = 2, where m s is the mass of the material (g) and V is the solution volume (L).
The statistical adequacy of the constructed models in itself does not guarantee their correctness.It is necessary to confirm the composition of fixed complexes and to verify the accuracy of the calculated fitting parameters.
To corroborate the binding of metal ions in the form of anionic chloro complexes, the spectral data have been involved.
As a result of the copper(II) adsorption from aqueous solutions, two clearly defined charge transfer peaks at 295 and 410 nm appear in the visible spectrum of the SMICl material (Figure 6a).They are unambiguously attributed to the unsolvated CuCl 4 2-anions, 22,[28][29][30][31][32] that confirms definitely the formation of only one surface complex (Cat + ) 2 CuCl 4 2-.The simulation results indicate that species containing CuCl 4 2-and CuCl 3 -anions coexist at the surface when CuCl 2 is adsorbed from alcohol media.The visible spectrum of the material with copper(II) species sorbed from alcohol solutions is split up into three Lorentzian functions with maxima at 325, 390 and 435 nm (Figure 6b).First two maxima can be interpreted as the overlapping absorption bands of complex anions CuCl 4 2-and CuCl 3 -, 22,[28][29][30] while the absorption at 435 nm is the distinctive feature of the CuCl 3 complex anion in which the fourth position is occupied by the alcohol molecule. 22,31n the electronic spectrum of the SMICl material with fixed cobalt(II) species (Figure 7) four absorption bands have been detected.Available literature on the assignment of maxima to the particular complexes is rather contradictive. 33,34However, the absorption band at ca. 620 nm is considered as the distinctive feature of the CoCl 4 2-complex, while the band at 580 nm is attributed definitely to the CoCl 3 -ion. 31The bands at 680 and 715 nm can be ascribed both to tri-and tetrachloro complexes. 21,31,33ote that wavelengths of the d-d transitions in chlorometallates are close to the infrared absorption maxima of imidazolium cation.Accordingly, the data on absorbance in the NIR region were not used in discussing the composition of fixed metal complexes.
The observed stretching frequencies (n(M-Cl)) in the Raman spectra (Table 3) are assigned to the vibrational A 1 mode of the tetrahedral chlorometallates. 35This confirms that cobalt(II), zinc(II), cadmium(II) and mercury(II) chlorides are fixed by the SMICl material in the form of higher chloro complexes and falls into line with the results of simulation of adsorption equilibria.
Judging by the effective sorption capacities of the material and the values of the equilibrium constants (Tables 2 and 4), it may be concluded that among the d 10 -ion chlorides the affinity of the material to metal chlorides rises in the order ZnCl 2 < CdCl 2 < HgCl 2 , i.e., in the same order in which the stability constants of metal-chloride complexes in solutions increase.
To check the reliability of the found fitting model parameters, some of experimental data were handled with the Langmuir adsorption model.As was mentioned above, this model is applicable for the systems in which fixed complexes of only one type are formed and dissociation of metal chlorides in solutions is negligible.These conditions are valid for ZnCl 2 and CdCl 2 adsorption from ethanol solutions (dissociation of metal chlorides does not exceed several percents, only fixed complexes containing MCl 3 -anions are formed) and for HgCl 2 sorption from methanol solutions (the salt does not dissociate, the yield of surface complex (Cat + ) 2 (HgCl 3 -) 2 is at least eight times higher than the yield of other complexes).The unambiguous relationship between σ 1 (2) and K L (1) valid for these conditions was derived: 10,37 A good agreement between the K L (1) values determined from the adsorption isotherms and recalculated by equation 13 (Table 4) corroborates the trustworthiness of the simulation results.
A deeper insight into the binding properties of materials may be gained via the comparison of the equilibrium constants obtained in this work with that for other ion exchangers described in the literature. 1,2,4,5,7,8able 5 presents the equilibrium constants for the CoCl 2 and CuCl 2 adsorption from ethanol solutions by several ion exchange materials containing Cl -counterions.The constants have been calculated proceeding from the earlier obtained experimental data with the account of the formation of several metal-chloride complexes in solutions.It is worth emphasizing that the above conclusions about the composition of fixed metal complexes remain unchanged.According to Tables 2 and 5 the constants of adsorption equilibria are rather close for different ion exchange materials.This allows to assert that the nature of immobilized organic cation and the properties of carrier are Here and below points represent experimental data, solid lines represent the Lorentzian (Figure 6) or Gaussian (Figure 7) fits, and the dash lines represent the fitted peaks.The frequencies assigned to the methylimidazolium cation correspond to those reported in the literature. 36Signal intensities: vw: very weak, w: weak, m: moderate, s: strong, vs: very strong. of lesser significance than the fact of metal fixation in the form of anionic chloro complexes.
To discuss the influence of solvent on the formation of fixed metal complexes it is necessary to take into account not only the constants of chemisorption equilibria, but also the effective sorption capacities.For salts that dissociate in solutions faintly, no significant differences between solvents have been observed.At the same time, the sorption capacity for CuCl 2 and CoCl 2 chemisorption from methanol solutions exceeds this characteristic for ethanol solutions.This feature can be ascribed to the higher ion solvation ability of methanol due to which the formation of metal chloride complexes in solutions competes with the fixation of metal ions more distinctly.
An advantage of the suggested approach to the description of adsorption equilibria is the possibility to use the obtained results for calculating the specific concentrations of fixed metal-chloride complexes dependently of the solvent, material dosage and initial concentration of metal chloride.
To efficiently use the SMICl material for the removal of metal chlorides from alcohol solutions, conditions that provide high degrees of metal extraction (ξ) need to be known.The calculation of equilibrium compositions of adsorption systems has allowed to estimate the influence of the material dosage on the metal removal and has shown that high ξ values can be easily reached.For example, 2 g of the SMICl material remove 75-90% of metal chlorides from 50 mL of solutions with the concentration 1.0 × 10 -3 mol L -1 .
The regeneration of the material is simple and consists in the short-term stirring of its aqueous suspension.This results in the release of a metal chloride into the aqueous phase.When the ratio adsorbent mass/solution volume is 0.2 g L -1 , 90-99% of a metal chloride is leached by the very first portion of water.To ensure the practically total removal of metal species, this procedure may be repeated several times.The acidification of solutions negligibly influence the efficiency of leaching.This is for the reason that the pH decrease below ca.5-5.5 does not affect the state of both fixed species and metal species in solution.

Conclusions
Silica modified with 1-n-propyl-3-methylimidazolium chloride adsorbs cobalt(II), copper(II), zinc(II), cadmium(II) and mercury(II) chlorides from methanol, ethanol and aqueous solutions in the form of metal-chloride anionic counterions.They are formed in the course of adsorption process via the chemical complexation reaction of a metal chloride from the liquid phase with chloride counterions of the adsorbent.This assertion follows from the simulation of adsorption equilibria and the data of electronic and Raman spectroscopy.The model of bidentate binding was used for the detailed analysis of adsorption equilibria.This model takes into account the dissociation and auto-complexation of metal chlorides in solutions as well as the possibility of formation of several fixed metal-chloride anionic complexes detained near the charged surface groups as counterions.It describes the results of measurements within the experimental errors and is adequate from the statistical point of view.The reliability of the calculated fitting parameters has been confirmed by the fact that the estimates determined in different ways coincide within the limits of their errors.The effective sorption capacities of the material bearing 1.0 mmol g -1 of the modifier towards metal chlorides vary from 0.06 to 0.50 mmol g -1 depending on particular solvent and metal chloride.The determined constants of adsorption equilibria manifest that the material removes efficiently all studied metal chlorides from alcoholic solutions and HgCl 2 from aqueous solutions.The model of bidentate binding that also accounts in details the metalchloride complexation equilibria in solutions was used to obtain more accurate information about the cobalt(II) and copper(II) chlorides adsorption from ethanol by several ion exchange materials with immobilized nitrogen containing cations and chloride counterions.The obtained constants of adsorption equilibria are rather close for different ion exchange materials.This suggests that the formation of metal-chloride anionic complexes rather than the nature of fixed organic cations or properties of carriers is the determinative factor of metal chlorides binding.

Figure 2 .
Figure 2. Variant of the model of bidentate binding.

Figure 4 .
Figure 4. Degrees of formation of species in the systems containing the SMICl material and CoCl 2 or CuCl 2 ethanol solutions.Here and in Figure 5 masses of the material are 0.05 g, volumes of solutions are 0.05 L, a M(II)chemisorbed = a (Cat + ) 2 MCl 42-+ a (Cat + ) 2 (MCl 3 -) 2 , (Cat + ) 2 (Cl -) 2 are unoccupied sites that represent the unreacted Cl -counterions at the material surface.

%Figure 5 .
Figure 5. Degrees of formation of species in the systems containing the SMICl material and CoCl 2 or CuCl 2 methanol solutions.

Figure 6 .
Figure 6.Absorption spectra of the SMICl material with copper(II) adsorbed from (a) aqueous and (b) ethanol solutions.N f (Cu II ) are 0.1 mmol g -1 (a) and 0.065 mmol g -1 (b).Here and below points represent experimental data, solid lines represent the Lorentzian (Figure6) or Gaussian (Figure7) fits, and the dash lines represent the fitted peaks.

Table 1 .
Logarithms of the overall (β i ) and stepwise (κ i ) stability constants of metal-chloride complexes in solutions at 298 K

Table 2 .
The models of adsorption equilibria

Table 3 .
The Raman frequencies for the SMICl material

Table 5 .
Results of simulation of CoCl 2 and CuCl 2 adsorption from ethanol solutions by various ion exchange materials

Table 4 .
Comparison of fitting parameters estimated in different ways