- Email: [email protected]
Adsorption of fluoride, chloride, bromide, and bromate ions on a novel ion exchanger
Journal of Colloid and Interface Science 291 (2005) 67–74 www.elsevier.com/locate/jcis
Adsorption of fluoride, chloride, bromide, and bromate ions on a novel ion exchanger N.I. Chubar a,∗ , V.F. Samanidou b , V.S. Kouts a , G.G. Gallios b , V.A. Kanibolotsky a , V.V. Strelko a , I.Z. Zhuravlev a a Institute for Sorption and Problems of Endoecology (ISPE), National Academy of Sciences of Ukraine, 13, General Naumov str., 03680 Kiev, Ukraine b Department of Chemistry, Aristotle University of Thessaloniki, GR-54124 Thessaloniki, Greece
Received 31 January 2005; accepted 27 April 2005 Available online 17 June 2005
Abstract A novel ion exchanger based on double hydrous oxide (Fe2 O3 ·Al2 O3 ·xH2 O) was obtained by the original sol–gel method from easily available and cheap raw materials and employed for adsorption of F− , Cl− , Br− , and BrO− 3 from simultaneous solutions. Adsorbent was characterized by potentiometric titration, ζ -potential, and poremetrical characteristics. A technologically attractive pH effect of F− , Br− , and BrO− 3 sorption on the investigated double hydroxide of Fe and Al, which is capable of working in the pH range 3 to 8.5, was observed. Kinetic data on fluoride and bromide sorption fit well the pseudo-second-order model. Isotherms of fluoride, bromide, chlorine, and bromate ion sorption on Fe2 O3 ·Al2 O3 ·xH2 O were obtained at pH 4. The isotherm of F− sorption fit well the Langmuir model; sorption affinity (K = 0.52 L/mg) and sorption capacity (90 mg F/g) were high. In the competitive adsorption of bromide and bromate, bromide dominated at equilibrium concentrations of the ions >40 mg/L. The mechanism of fluoride adsorption to the surface of the model cluster of the sorbent synthesized and the geometry of the cluster itself were modeled with the HyperChem7 program using the PM3 method. 2005 Elsevier Inc. All rights reserved. Keywords: Inorganic ion exchangers; Adsorption; Anions; Isotherms; Quantum chemistry modeling
1. Introduction The anions (F− , Br− , Cl− , and BrO− 3 ) were chosen for study because of their harmful effects on human health when present in large concentrations. Fluoride is an essential micronutrient for living beings. The optimum fluoride level in drinking water set by WHO is between 0.5 and 1.0 mg/L (http://www.who.int). Concentrations higher than this level can lead to fluorosis [1], caused by excess ingestion of fluoride. Severe forms of the disease typically develop only when the F− concentration of drinking water is greater than 5–10 mg/L. Wastewater from phosphate fertilizer plants may contain up to 2% of fluoride [2]. Increased levels of fluoride can also be present in effluents from the fluorine in* Corresponding author. Fax: +38 044 452 93 27.
E-mail address: [email protected] (N.I. Chubar). 0021-9797/$ – see front matter 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.jcis.2005.04.086
dustry, from glass etching [2], and in groundwater around aluminum smelters [3,4]. BrO− 3 is a powerful oxidant and has been shown to cause kidney and possibly other tumors in laboratory animals [5]. BrO− 3 does this directly or indirectly by causing oxidation of the lipids in cell membranes to produce active oxygen species, which in turn produce damage in macromolecules such as DNA. This anion is also genotoxic in vitro and in vivo, although this has been confirmed to cause primarily physical damage in chromosomes [6]. Chlorination is a cornerstone of water treatment that produces potential carcinogens, such as the trihalomethanes, as disinfection byproducts (DPBs). Ozonation is one of the most promising alternatives to chlorination because it avoids the formation of dangerous DPBs and, in addition, is very effective in removing pesticides and other hazardous water components [7]. However, despite the advantages of ozonation, the main dis-
68
N.I. Chubar et al. / Journal of Colloid and Interface Science 291 (2005) 67–74
advantage is formation of bromate ions. WHO has proposed a guideline of <0.5 ppb bromate ions in drinking water. It is obvious that one of the main ways to reduce bromate concentration is to remove bromide ions from the raw water before ozonation [8]. Removal of bromide has particular significance with respect to obtaining control over formation of brominated DBPs. For example, observations in a number of regions indicate that a reduction in natural organic matter (NOM) prior to chlorination does not result in a corresponding reduction in brominated DBPs if the bromide levels are not reduced as well [9]. This can produce conditions that limit the extent to which reductions in total trihalomethane formation can be achieved. Therefore, the ability to remove bromide would be an additional asset in meeting regulatory requirements. It has also been found that chloride and bromide ions deeply inhibit the rate of degradation of chloroform and tetrachloromethane [10]. The major techniques for removal of fluoride and bromide ions on a large scale include precipitation and coagulation, as well as adsorption and ion exchange. The viability of the last method is dependent completely on the development of adsorptive materials. In contrast to the materials widely used for anion removal (activated alumina [11,12], activated carbon [13,14], and ion-exchange resins [15]), inorganic ion exchangers are considered the most prospective because of their chemical stability and possible ability to control surface chemistry [16–19]. Traditionally, inorganic adsorbents have been obtained by hydrolysis (precipitation) of salts. Recently, a sol–gel method based on controlled hydrolysis of expensive alkoxy compounds of the corresponding metals has been used more extensively. At the Institute for Sorption and Problems of Endoecology, National Academy of Sciences of Ukraine, an original sol–gel method of synthesis has been developed [20,21] that allows production of spherically granulated sorbents from inexpensive raw materials (simple inorganic salts of the corresponding metals, mineral acids, and alkalis). The main objectives of this work were to use the sol–gel method to synthesize a novel inorganic ion exchanger based on a mixed hydrous oxide of iron and aluminum, to investigate adsorption of fluoride, chloride, bromide, and bromate ions under batch conditions (pH effect, isotherms, and kinetics), and to use the quantum chemistry HyperChem7 program (PM3 method) to optimize the model cluster of the sorbent obtained and propose the mechanism of fluoride adsorption.
Table 1 Texture characteristics of Fe2 O3 ·Al2 O3 ·xH2 O Formula (mol) Specific surface area, Ss (m2 /g) Pore volume, Vs (cm3 /g) On water On benzene Moisture (mass%)
Fe1.0 ·Al0.93 ·(H2 O)1.23 396 0.15 0.28 20.4
used were of analytical grade. The raw materials (mixture of the initial salts) were subjected first to preliminary neutralization with a water solution of ammonia. The streams of these precursors (sols of Al–Fe–(OH)5 Cl) were dispersed dropwise [34] in undecane, and the small drops formed in undecane, fell down into an ammonia solution, where sols were transferring into gels (hardening), forming spherical granules with little deformation. The granules were washed from the NH4 Cl with distilled water, treated hydrothermally in an autoclave at 90–95 ◦ C for 6 h and dried in air at room temperature for 48 h. The sorbent was produced according to the general reaction scheme FeCl3 + AlCl3 + 6NH4 OH → Fe(OH)3 ·Al(OH)3 + 6NH4 Cl. X-ray analysis confirmed that the material was amorphous. Its texture characteristics are listed in Table 1. Specific surface area was determined by the single-point N2 -BET method [22]. Pore volume on water and benzene and moisture content were determined as described in Ref. [23]. Note that the sorbent has a high surface area. For comparison, goethite, which has also been used for anion adsorption, has a specific surface area of 33.5 m2 /g [24]; however, there are some materials with areas much higher than 33.5 m2 /g. Particle size of the adsorbents investigated was 0.5–1.0 mm. 2.2. Potentiometric titration The cation and anion exchange capacity of the investigated sorbents (with respect to H+ and OH− ions correspondingly) was studied by potentiometric titration as described [25,26]. Typically, 1 g of ion exchanger was introduced into a flask containing 50 cm3 of 0.1 M NaCl; then a definite volume of 0.1 M NaOH or HCl was added, the flask was placed in a temperature-controlled orbital shaker for 48 h, and equilibrium pH was measured. Ion exchange capacity, q (mEq/g) was calculated using the equation C(V1 − V2 ) , 1000 × m
2. Materials and methods
q=
2.1. Synthesis of the sorbent and its texture characteristics
where C (mol/ml) is the concentration of the acid/alkali, V1 (ml) is the volume of added NaOH/HCl in the experiment at defined pH, V2 (ml) is the volume of NaOH/HCl added in the blank experiment at the same pH, and m (mg) is the mass of the sorbent.
Chloride salts of iron and aluminum (FeCl3 ·6H2 O and AlCl3 ·6H2 O) and a water solution of ammonia (25%) were used to synthesize the novel ion exchangers. All reagents
(1)
N.I. Chubar et al. / Journal of Colloid and Interface Science 291 (2005) 67–74
2.3. ζ -Potential studies ζ -Potential measurements were carried out with a Microelectrophoretic zeta potential analyzer (Mark II produced by Rank Brothers Co.). The samples were previously prepared as follows: Two hundred milligrams of the sorbent was added to the flask containing 100 ml of water, the pH of the solution was adjusted with 0.1 M HCl or NaOH solution, and the flask was placed in a temperature-controlled orbital shaker for 48 h at 22 ± 2 ◦ C. The electrophoretic mobility of the dispersion transferred into the microelectrophoretric cell was measured. The ζ potential was calculated using the Smoluchovski equation [27,28]. 2.4. Batch sorption experiment Experiments were carried out in Erlenmeyer flasks placed in a temperature-controlled orbital shaker (stirring speed of 150 min−1 ) for 48 h as it had been established previously that 48 h is equilibrium time. Two hundred milligrams of the sorbent was equilibrated to 100 ml of the anion solution at 22 ± 2 ◦ C using the background electrolyte 0.01 M NaNO3 . It was shown separately that NO− 3 is not adsorbed by the surface of this ion exchanger and cannot compete with halides for sorption sites. A suitable volume of acid (HNO3 ) or base (NaOH) solution was added to adjust pH, which was measured with a digital pH meter (TOA Electronics Ltd., HM-35 V). The adsorbent was removed by centrifugation, and the concentration of the corresponding anion remaining in the supernatant was determined by single-column ion chromatography with a low-conductivity mobile phase. The latter consisted of a mixture of 4 mM p-hydroxybenzoic acid and 0.5 mM sodium benzoate (pH 9.0 adjusted with 1 M NaOH) delivered at the flow rate of 2.3 ml/min. The ion chromatograph consisted of an SSI 222D Pump (SSI, State Colleague, PA, USA) and a Wescan 315 conductivity detector (Alltech, Deerfield, IL, USA) maintained at 25 ± 0.1 ◦ C. A Rheodyne 9125 six-port highpressure switching injection valve (Rheodyne, Cotati, CA, USA) was assembled with a 50-µl injection loop. The separation column Hamilton PRP-X100, 150 × 4.1 mm (Hamilton, Reno, NV, USA), was packed with spherical 10-µm polystyrene–divinylbenzene trimethylbenzene trimethylammonium anion exchanger (Hamilton, Reno, NV, USA). The capacity of the column was 0.19 ± 0.02 mEq/g. Data acquisition was performed using a Hewlett–Packard integrator (Model HP 3396 II, Hewlett–Packard, Avondale, PA, USA). The final pH of the solutions was also measured. The experiments were repeated two or three times and the average values were used for data treatment. Anion uptake was determined according to the equation (C0 − Ceq )V , (2) M where q (mg/g) is the amount of anion adsorbed per gram of oxide hydrated, C0 (mg/ml) is the initial concentration of q=
69
the anion in solution, Ceq (mg/ml) is the final (or equilibrium) concentration of the anion in solution, V (ml) is the volume of the solution, and M (g) is the mass of the sorbent.
2.5. Fitting the isotherms to the Langmuir model To describe the mechanism of sorption and to draw up the affinity constants, the experimental results were fit to the linear form by the Langmuir equation, Ceq Ceq 1 = + , qeq Kqmax qmax
(3)
where Ceq (mg/dm3 ) is the equilibrium concentration of the anion in solution, qeq (mg/g) is the correspondent capacity of the sorbent, qmax (mg/g) is the maximum anion concentration sorbed by the ion exchanger, and K (dm3 mg−1 ) is the affinity constant [27,28]. 2.6. Kinetics of the sorption of the investigated anion The kinetics of fluoride and bromide ion sorption was fitted well to a pseudo-second-order mechanism corresponding to the equation dy (4) = Kc (1 − y)2 , dt where Kc is the kinetic constant. Solving this ordinary differential equation alone with the conditions that t = 0, y = 0, Eq. (6) can be linearized into the form, 1 t 1 = + t, q Kc qe qe
(5)
which allows calculation of qe and Kc , respectively, from the slope and intercept of the straight line. Ho [29] was the first to develop the pseudo-second-order kinetic expression for sorption systems, in which chemical sorption is the rate-limiting step; since its development this model has been applied in a number of studies [30–32] including those for sorption of anions [32]. The pseudo-first-order model developed by Lagergren (1998) did not work as well in our experimental data. That is why we described the data with a pseudo-second-order mechanism.
3. Results and discussion 3.1. Surface studies of Fe2 O3 ·Al2 O3 ·xH2 O It is known that hydrated metal oxides are ampholites with cation and anion exchange properties. That is why the materials investigated have shown both cation and anion exchange capacity, which reached values of 1.38 and 1.8 mEq/g, respectively (Fig. 1a). The isoelectric point of the sorbents was at neutral pH.
70
N.I. Chubar et al. / Journal of Colloid and Interface Science 291 (2005) 67–74
(a)
Fig. 2. Plot of sorption of F− , Br− , Cl− , and BrO− 3 versus pH on Fe2 O3 · Al2 O3 ·xH2 O. Conditions of the experiment: 48-h shaking time, 20 ± 2 ◦ C, 120 mg/L initial concentration of each ion, 0.01 M NaNO3 background electrolyte, and 200 mg/L solid concentration.
(b) Fig. 1. Potentiometric titration curve (a) and effect of pH on ζ potential of Fe2 O3 ·Al2 O3 ·xH2 O at background electrolyte 0.01 M NaNO3 (b).
The effect of pH on ζ potential on hydrous oxides of Fe and Al is illustrated in Fig. 1b. The point of zero charge was at pH 3.6. The surface of the materials was negatively charged at pH higher than 3.6. The maximum negative charge was at pH 9. 3.2. Influence of pH on anion sorption Fig. 2 shows the pH effect of anion sorption on Fe2 O3 · Al2 O3 ·xH2 O (mentioned as Fe–Al); the initial concentration of all ions was 120 mg/L. It was found that the sorption capacity of F− on the double hydrous oxide investigated was higher compared with that of the other halides. The higher affinity of the sorbent to fluoride is due to possible formation of insoluble iron fluoride and less soluble aluminum fluorides. The technologically important (for drinking water treatment) pH effect on fluoride, bromide, and bromate sorption on the adsorbent investigated was established. Typically, sorption of anions on individual hydrous oxides is very dependent on pH. For example, F− sorption on the natural mineral goethite [32] and acid-treated spent bleaching earth [33] takes place in acidic solutions only. Sorption of F− decreases slowly with increasing pH, and the anions are not adsorbed at all at pH 7. Fe–Al could take up F− (as well as Br− and BrO− 3 ) (Fig. 2) in the pH range 3–8.5. In acidic solutions at pH 3–4, the sorption capacity of the adsorbent investigated toward fluoride was twice as high as in neutral and alkaline solutions. At pH 4–5 a sharp decrease in sorption from 70 to 37 mg/g takes place. At pH 7.5–8 some decrease in the sorption capacity of the sorbent (near 20 mg/g) occurs. pH < 5 is the most favorable for fluoride uptake due to the lower concentration of OH− ions, which
compete with fluoride for sorption sites with increasing concentration of OH− in the solution and reduce the capacity for sorption of F− . Speciation of the investigated anion is the second reason for higher fluoride sorption at pH < 5 when it exists as HF. Maximal sorption capacity is at pH 4, at which this is the predominant form of the anion. The adsorption of F− can be described as an exchange reaction against OH of the surface groups. IR analysis by Hiemstra and colleagues [32], who investigated sorption of F− on goethite, confirmed that the main reaction of such an adsorption process is singly coordinated FeOH groups. They formulated the formation of FeF−1/2 as FeOH−1/2 + H+ (aq) + F− (aq) FeF−1/2 + H2 O.
(6)
Fluoride species at pH < 5 (HF) are the easiest adsorbing due to formation of water molecules in accordance with the reaction [1]. Sorption methods are not widely accepted for removal of Cl− . We tried to assess the sorption of this anion on Fe–Al to compare the data with data for other halides. Fig. 2 shows that sorption of Cl− on Fe–Al depends very much on pH, and this anion can be adsorbed at pH 3–4 followed by sharp degrees of sorption that become zero at pH 5. This can be explained: ion exchange is the main mechanism for chloride adsorption on the materials investigated and Cl− cannot compete with increasing OH− concentration at higher pH. The pH effect of Br− sorption was similar to the pH effect of F− sorption. Maximum bromide ion sorption also occurred at pH < 5 due to the lower concentration of OH− ions, which compete for sorption sites, and the predominant existence of fluoride as HBr species. Sorption capacity then decreased slowly and reached a small, second peak at pH 9. Sorption of bromate ions was approximately equal through the whole pH interval investigated (3–10), with some small maxima at pH 3 and 9–10. This anion has another mechanism of sorption (confirmed by isotherms, see Fig. 8) which should be studied separately.
N.I. Chubar et al. / Journal of Colloid and Interface Science 291 (2005) 67–74
71
Fig. 3. Scheme of the model cluster of 4Fe–4Al.
3.3. Modeling of the sorption mechanism using quantum-chemistry program The ability of the sorbent to take up F− , Br− , and BrO− 3 ions at pH > 5 can be explained by the fact that pH complexation plays an important role in the binding of these anions to the surface. Most probably, the anions formed complexes with Al atoms due to the higher stability of AlF2+ , AlF+ 2, and AlF3 (pK 6.13, 5.02, and 3.83, respectively) as compared with the stability of FeF2+ , FeF+ 2 , and FeF3 (pK 5.17, 3.83, and 2.91, respectively) [34]. Additional evidence of this suggestion was drawn from some modeling of the adsorbent (4Fe–4Al) structure with HyperChem7 (Figs. 3 and 4). Fig. 3 shows the model cluster of the synthesized sorbent containing equal numbers of iron and aluminum atoms. The electronic structure of this cluster (geometry and charge distribution) has been calculated using the PM3 method. In this model cluster, an atom of aluminum is 3-coordinated, whereas an atom of iron has octahedral and tetrahedral environments. Oxygen is 2- and 3-coordinated here. The charges on atoms of aluminum, iron, and also oxygen of hydroxyl groups linked with aluminum and iron have been shown. There is a large positive charge (near +e) on the aluminum atom, but a small negative charge on the iron atoms. This proves that electronic density has been transfered from oxygen atoms linked with iron. Analysis of the local functions (on atoms) of the density of states (DOS) has shown that the atoms of oxygen and iron have considerable electron-donor properties, whereas aluminum and oxygen have electron-acceptor properties. It is noteworthy that octahedral atoms of iron have more electronacceptor properties but tetrahedral iron has more electrondonor properties. Thus , 3-coordinated atoms of oxygen have the highest electron-donor and electron-acceptor properties. Oxygen atoms of hydroxyl groups are not responsible for electron-acceptor properties.
Fig. 4. Local functions of the density of states (DOS) of cluster 4Fe–4Al.
The charge distribution on the surface, shown in Fig. 1, indicates that considerable positive charges, concentrated on aluminum, are the most probable (reasonable) coordination sites for negatively charged halides. The mechanism of this reaction has been demonstrated by interaction of NaF with the model cluster investigated (Fig. 3). Taking into account that dipole moments of the cluster and NaF are correspondingly 7.07 and 8.11 D, electrostatic interactions, which cause the molecule NaF to orientate to and approach the atom of Al having the biggest charge (q = 1.117e) on the cluster, will dominate at considerable distances between NaF and the cluster (1, 2, and 3 in Fig. 5). Optimization of the geometry of the cluster (3, Fig. 5) by the PM3 method allows us to conclude that during interaction, a transition reactive complex with 4-coordinated Al (AlO3 F) is formed. Such an interaction leads to redistribution of the charge density, separation of Na+ ions, and formation of hydrated cations of Na+ , which interact with OH− of 4-coordinated Al to form NaOH (5 and 6 in Fig. 5). Isolation of OH− and alkalization of the experimental solution are also confirmed by the experimental data. A final result of this interaction is a cluster with bound fluoride. It is worth noting that gain in heat formation of the F-containing cluster as compared with the initial cluster is 37 kcal/mol. Formation of clusters with bound Cl and Br leads to decrease in heat formation (49 and 36 kcal/mol). This means that the binding of fluoride to the surface of the cluster investigated occurs much more easily than the binding of Br and Cl, which is in good correlation with the experimental data in this article.
72
N.I. Chubar et al. / Journal of Colloid and Interface Science 291 (2005) 67–74
Fig. 5. Mechanism of fluoride binding to the surface of 4Fe–4Al.
3.4. Sorption kinetics Fluoride and bromide sorption was found to be relatively fast kinetically. All fluoride ions present in the solution (200 mg/L) with buffer background electrolyte were adsorbed for 20 min. In the first 10 min 50% of the bromide ions and 75% of the fluoride ions were adsorbed (Fig. 6). The kinetics data fit well to the pseudo-second-order model. The rate coefficients of fluoride and bromide sorption calculated from the model were 0.40 and 0.16 min−1 , respectively. Sorption of F− was 2.5 times faster than sorption of Br− under the same conditions. For comparison, adsorption of F− on the ion exchanger investigated was slower than on new ion exchange fiber [15] but slightly faster than on treated spent bleaching earth [33].
(a)
3.5. Isotherms The isotherm of fluoride sorption (Fig. 7), which is very close to the y axis as compared with the curves obtained for the other anions investigated in this article, is evidence of the high affinity of the adsorbent investigated to F− . The capacity of the ion exchanger for sorption of F− (4.2 mmol/g) was four times higher as compared with the sorption capacity of the new ion exchange fiber [15] and 10 times higher than that of goethite [32]. Hiemstra and colleagues [32] studied the mechanism of F− sorption using the mathematical modeling approach CD-MUSIC. We suppose that similar processes occur on Fe–Al. It was concluded that at the lower concentration of F− in the solution, these ions were exchanged with OH− , coordinated on the surface, in a one-charge scheme.
(b) Fig. 6. Kinetics of F− and Br− sorption (a) on Fe2 O3 ·Al2 O3 ·xH2 O and fitting of the sorption kinetics to the pseudo-second-order model (b). Conditions of the experiment: initial concentration 200 mg/L; 0.01 M NaNO3 background electrolyte, pH 4; 20 ± 2 ◦ C.
At higher concentrations of fluoride ions, OH− ions, coordinated on the two-point scheme, take part in the ion exchange with F− . Redistribution of the surface charge takes place
N.I. Chubar et al. / Journal of Colloid and Interface Science 291 (2005) 67–74
(a)
73
− Fig. 8. Isotherms of Br− , BrO− 3 , and Cl sorption on Fe2 O3 ·Al2 O3 ·xH2 O. Conditions of the experiment: 0.01 M NaNO3 background electrolyte, pH 4, 48-h shaking time; 20 ± 2 ◦ C.
(b) Fig. 7. Isotherms of F− sorption on Fe2 O3 ·Al2 O3 ·xH2 O and fitting of the data of curve (a) to the Langmuir model (b). Background electrolyte 0.01 M NaNO3 , pH 4.
on the formation of surface complexes of fluoride. This is explained by a change from donor-type complexes (FeOH, Fe(OH)2 ) to proton-acceptor complexes (FeF3 ). Distribution of the charge in this model is approximately equivalent to the change from donor H bond to acceptor H bond. At very high concentrations of fluoride ions, precipitation of F− into FeF3 occurred. The high concentrations of background electrolytes catalyze this process. Isotherms of F− sorption fit well the Langmuir model with a great correlation coefficient of 0.9987. Maximum sorption capacity calculated from the model was close to the experimental data (90 mg/g), and the affinity constant was 0.52 L/mg, which is 10 times higher than analogous parameters measured for acid-treated spent bleaching earth [33]. Fig. 8 shows the sorption isotherms of Br− , Cl− , and BrO− 3 . The character of these curves is evidence of the lower affinity of the sorbents to these anions, as compared with the close-to-axis y isotherm of F− sorption. Sorption isotherms of BrO− 3 have two plateaus at the equilibrium concentrations of the ions: 20–60 and >90 mg/L. Sorption capacity was 15 and 25 mg/g. Growth of sorption was noted at equilibrium concentrations from 60 to 80 mg/L; then, the second plateaus were found. The difference in the form of isotherms of F− and Br− , on one side, and BrO− 3 , on the other side, can be explained by their different affinities to the anion, different mechanisms of sorption, and speciation. The first plateaus on the curve meant that the most available sorption
Fig. 9. Isotherms of competitive adsorption of Br− and BrO− 3 onto Fe2 O3 · Al2 O3 ·xH2 O under the conditions of their combined presence in the solution at the same concentrations (0.01 M NaNO3 background electrolyte, 48-h shaking time).
sites had been saturated. At the second plateaus, adsorption was taking place on the less favorable sites. Adsorption of bromide was higher as compared with adsorption of bromate ions as confirmed by the study of the competitive adsorption of these two ions (Fig. 9). Investigation of the competitive adsorption of Br− and BrO− 3 added in equal concentrations to experimental solutions has shown that at lower concentrations of the anions investigated (Ceq < 40 mg/L), bromate dominated in the competitive adsorption with bromide ions. At higher concentrations of the anions (>70 mg/L), on the contrary, sorption of BrO− 3 was already zero and Br− was the only ion adsorbed to the surface of the ion exchanger investigated. Such data on competitive adsorption are in good agreement with isotherms for each anion adsorption shown in Fig. 8.
4. Conclusions The anion exchange capacity of the novel inorganic ionite based on the mixed oxide hydrate of iron and aluminum (equivalent ratio) was 1.8 mEq/g, in accordance with the potentiometric titration data. The pH effect of F− and Br− was dependent on ion speciation; however, sufficient capacity for sorption of these anions was observed in the pH
74
N.I. Chubar et al. / Journal of Colloid and Interface Science 291 (2005) 67–74
range 3–10. Adsorption of fluoride (88 mg/g) was the highest among the anions investigated and fit well the Langmuir model with a high affinity constant. The kinetics of F− and Br− sorption was subject to the pseudo-second-order model with rate coefficients of 0.40 and 0.16 min−1 , respectively. Isotherms for the competitive adsorption of Br− and BrO− 3 were in the agreement with isotherms for each ion: at lower concentrations of the ions in the solution (<40 mg/L), bromate dominated; then its sorption was depressed completely by bromide ions. The model cluster of 4Fe–4Al was optimized with HyperChem7 (PM3 method) and the charge distribution on the cluster was obtained. Analysis of the local (on atoms) functions of the state density has shown that 3-coordinated oxygen atoms have the highest electron-donor and electron-acceptor properties. Modeling of the NaF interaction with the model cluster surface has shown that binding of F takes place on the aluminum atoms and this interaction is energetically favorable (gain of heat formation) for fluoride but not for bromide and chloride ion binding.
Acknowledgments Work was supported by a NATO grant to Dr. N. Chubar, Aristotle University, Thessaloniki (Greece), by STCU Collaborative Grant N1739 (manager: Professor V.V. Strelko), and by Bilateral Collaborative Grant M/275 (manager: N. Chubar).
References [1] B.S. Zhang, Research on Endemic Fluoride Toxicosis and Industrial Fluoride Pollution, Chinese Environmental Sciences Press, Beijing, 1994. [2] D. Ró˙zycka, M. Grobelny, Chemik 4 (1998) 92–93. [3] S.K. Sahu, S.S. Pati, R.K. Badapanda, Environ. Ecol. 16 (1998) 169– 171. [4] A.K.M. Arnesen, Water Air Soil Pollut. 103 (1998) 375–388. [5] Y. Kurokawa, A. Maekawa, M. Takahashi, Y. Kayashi, Environ. Health Perspect. 87 (1990) 309–335. [6] R.L. Melnick, Regul. Toxic. Pharmacol. 16 (1992) 111–125.
[7] J. Lawrence, H. Tosine, F.I. Onuska, M.E. Comba, Sci. Eng. 2 (1980) 55–64. [8] A. Mills, A. Belgrhazi, D. Rodman, Water Res. 30 (9) (1996) 1973– 1978. [9] J.G. Jacangelo, J. Am. Water Works Assoc. 87 (1995) 64–77. [10] P. Calza, E. Pelizzetti, Pure Appl. Chem. 73 (2001) 1839–1848. [11] J.H. Harmon, S.G. Kalichman, J. Am. Water Works Assoc. 57 (1965) 245. [12] E. Bellack, J. Am. Water Works Assoc. 63 (1971) 454. [13] D.S. Bhargava, S.B. Sheldarkar, Environ. Pollut. 76 (1992) 51. [14] M. Siddiqui, W. Zhai, G. Amy, C. Mysore, Water Res. 30 (1996) 1651–1660. [15] L. Ruixia, J. Guo, T.J. Hongxiao, J. Colloid Interface Sci. 248 (2002) 268. [16] J.M. DeMarco, A.K. SenGupta, J.E. Greenleaf, Water Res. 37 (2003) 164. [17] N. Gangoli, G. Thodos, J. Water Pollut. Control Fed. 45 (1973) 842. [18] S. Tokunaga, S.A. Wasay, S.W. Park, Water Sci. Technol. 35 (1997) 71. [19] S.A. Wasay, A. Tokunaga, S.W. Park, Sep. Sci. Technol. 31 (1996) 1501. [20] I.Z. Zhuravlev, V.A. Kanibolotsky, V.V. Strelko, A.I. Bortun, L.I. Bortun, S.A. Khainokov, A. Clearfield, J. Solv. Extract. Ion Exchange 17 (1999) 635. [21] A.I. Bortun, S.A. Khainokov, V.V. Strelko, I.A. Farbun, in: J.A. Grieg (Ed.), Proc. Ion Exchange Developments and Application, Cambridge, UK, 1996, pp. 305–310. [22] S. Brunauer, P.H. Emmeet, E. Teller, J. Am. Chem. Soc. 60 (1938) 309. [23] N.B. Keltsev, The Basis of Sorption Techniques, Chemistry, Moscow, 1984. [24] G. Yan, A. Mucci, Chem. Geol. 199 (2003) 91. [25] F. Helfrich, Ionites, Foreign Literature, Moscow, 1962. [26] R. Grissbach, Theory and Practice of the Ion Exchange, Foreign Literature, Moscow, 1963. [27] L. Svarovsky, Solid–Liquid Separation, third ed., Butterworths, London, 1990. [28] S.S. Voyutsky, Course of Colloidal Chemistry, Chemistry, Moscow, 1964. [29] Y.S. Ho, Ph.D. thesis, University of Birmingham, UK, 1995. [30] Y.S. Ho, G. McKay, J. Water Res. 34 (2000) 735. [31] R. Machado, J.M.R. Carvalho, M.J.N. Correia, J. Chem. Technol. Biotechnol. 77 (2002) 1340. [32] T. Hiemstra, W.H. Van Riemsdijk, J. Colloid Interface Sci. 225 (2000) 94–104. [33] M. Mahramanlioglu, I. Kizilcikli, I.O. Bicer, J. Fluoride Chem. 115 (2002) 41–47. [34] I.T. Goronovsky, Y.P. Nazarenko, E.F. Nekryach, in: O.D. Kurilenko (Ed.), Short Reference Book on Chemistry, Naukova Dumka, Kiev, 1974, p. 990.
Adsorption of fluoride, chloride, bromide, and bromate ions on a novel ion exchanger N.I. Chubar a,∗ , V.F. Samanidou b , V.S. Kouts a , G.G. Gallios b , V.A. Kanibolotsky a , V.V. Strelko a , I.Z. Zhuravlev a a Institute for Sorption and Problems of Endoecology (ISPE), National Academy of Sciences of Ukraine, 13, General Naumov str., 03680 Kiev, Ukraine b Department of Chemistry, Aristotle University of Thessaloniki, GR-54124 Thessaloniki, Greece
Received 31 January 2005; accepted 27 April 2005 Available online 17 June 2005
Abstract A novel ion exchanger based on double hydrous oxide (Fe2 O3 ·Al2 O3 ·xH2 O) was obtained by the original sol–gel method from easily available and cheap raw materials and employed for adsorption of F− , Cl− , Br− , and BrO− 3 from simultaneous solutions. Adsorbent was characterized by potentiometric titration, ζ -potential, and poremetrical characteristics. A technologically attractive pH effect of F− , Br− , and BrO− 3 sorption on the investigated double hydroxide of Fe and Al, which is capable of working in the pH range 3 to 8.5, was observed. Kinetic data on fluoride and bromide sorption fit well the pseudo-second-order model. Isotherms of fluoride, bromide, chlorine, and bromate ion sorption on Fe2 O3 ·Al2 O3 ·xH2 O were obtained at pH 4. The isotherm of F− sorption fit well the Langmuir model; sorption affinity (K = 0.52 L/mg) and sorption capacity (90 mg F/g) were high. In the competitive adsorption of bromide and bromate, bromide dominated at equilibrium concentrations of the ions >40 mg/L. The mechanism of fluoride adsorption to the surface of the model cluster of the sorbent synthesized and the geometry of the cluster itself were modeled with the HyperChem7 program using the PM3 method. 2005 Elsevier Inc. All rights reserved. Keywords: Inorganic ion exchangers; Adsorption; Anions; Isotherms; Quantum chemistry modeling
1. Introduction The anions (F− , Br− , Cl− , and BrO− 3 ) were chosen for study because of their harmful effects on human health when present in large concentrations. Fluoride is an essential micronutrient for living beings. The optimum fluoride level in drinking water set by WHO is between 0.5 and 1.0 mg/L (http://www.who.int). Concentrations higher than this level can lead to fluorosis [1], caused by excess ingestion of fluoride. Severe forms of the disease typically develop only when the F− concentration of drinking water is greater than 5–10 mg/L. Wastewater from phosphate fertilizer plants may contain up to 2% of fluoride [2]. Increased levels of fluoride can also be present in effluents from the fluorine in* Corresponding author. Fax: +38 044 452 93 27.
E-mail address: [email protected] (N.I. Chubar). 0021-9797/$ – see front matter 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.jcis.2005.04.086
dustry, from glass etching [2], and in groundwater around aluminum smelters [3,4]. BrO− 3 is a powerful oxidant and has been shown to cause kidney and possibly other tumors in laboratory animals [5]. BrO− 3 does this directly or indirectly by causing oxidation of the lipids in cell membranes to produce active oxygen species, which in turn produce damage in macromolecules such as DNA. This anion is also genotoxic in vitro and in vivo, although this has been confirmed to cause primarily physical damage in chromosomes [6]. Chlorination is a cornerstone of water treatment that produces potential carcinogens, such as the trihalomethanes, as disinfection byproducts (DPBs). Ozonation is one of the most promising alternatives to chlorination because it avoids the formation of dangerous DPBs and, in addition, is very effective in removing pesticides and other hazardous water components [7]. However, despite the advantages of ozonation, the main dis-
68
N.I. Chubar et al. / Journal of Colloid and Interface Science 291 (2005) 67–74
advantage is formation of bromate ions. WHO has proposed a guideline of <0.5 ppb bromate ions in drinking water. It is obvious that one of the main ways to reduce bromate concentration is to remove bromide ions from the raw water before ozonation [8]. Removal of bromide has particular significance with respect to obtaining control over formation of brominated DBPs. For example, observations in a number of regions indicate that a reduction in natural organic matter (NOM) prior to chlorination does not result in a corresponding reduction in brominated DBPs if the bromide levels are not reduced as well [9]. This can produce conditions that limit the extent to which reductions in total trihalomethane formation can be achieved. Therefore, the ability to remove bromide would be an additional asset in meeting regulatory requirements. It has also been found that chloride and bromide ions deeply inhibit the rate of degradation of chloroform and tetrachloromethane [10]. The major techniques for removal of fluoride and bromide ions on a large scale include precipitation and coagulation, as well as adsorption and ion exchange. The viability of the last method is dependent completely on the development of adsorptive materials. In contrast to the materials widely used for anion removal (activated alumina [11,12], activated carbon [13,14], and ion-exchange resins [15]), inorganic ion exchangers are considered the most prospective because of their chemical stability and possible ability to control surface chemistry [16–19]. Traditionally, inorganic adsorbents have been obtained by hydrolysis (precipitation) of salts. Recently, a sol–gel method based on controlled hydrolysis of expensive alkoxy compounds of the corresponding metals has been used more extensively. At the Institute for Sorption and Problems of Endoecology, National Academy of Sciences of Ukraine, an original sol–gel method of synthesis has been developed [20,21] that allows production of spherically granulated sorbents from inexpensive raw materials (simple inorganic salts of the corresponding metals, mineral acids, and alkalis). The main objectives of this work were to use the sol–gel method to synthesize a novel inorganic ion exchanger based on a mixed hydrous oxide of iron and aluminum, to investigate adsorption of fluoride, chloride, bromide, and bromate ions under batch conditions (pH effect, isotherms, and kinetics), and to use the quantum chemistry HyperChem7 program (PM3 method) to optimize the model cluster of the sorbent obtained and propose the mechanism of fluoride adsorption.
Table 1 Texture characteristics of Fe2 O3 ·Al2 O3 ·xH2 O Formula (mol) Specific surface area, Ss (m2 /g) Pore volume, Vs (cm3 /g) On water On benzene Moisture (mass%)
Fe1.0 ·Al0.93 ·(H2 O)1.23 396 0.15 0.28 20.4
used were of analytical grade. The raw materials (mixture of the initial salts) were subjected first to preliminary neutralization with a water solution of ammonia. The streams of these precursors (sols of Al–Fe–(OH)5 Cl) were dispersed dropwise [34] in undecane, and the small drops formed in undecane, fell down into an ammonia solution, where sols were transferring into gels (hardening), forming spherical granules with little deformation. The granules were washed from the NH4 Cl with distilled water, treated hydrothermally in an autoclave at 90–95 ◦ C for 6 h and dried in air at room temperature for 48 h. The sorbent was produced according to the general reaction scheme FeCl3 + AlCl3 + 6NH4 OH → Fe(OH)3 ·Al(OH)3 + 6NH4 Cl. X-ray analysis confirmed that the material was amorphous. Its texture characteristics are listed in Table 1. Specific surface area was determined by the single-point N2 -BET method [22]. Pore volume on water and benzene and moisture content were determined as described in Ref. [23]. Note that the sorbent has a high surface area. For comparison, goethite, which has also been used for anion adsorption, has a specific surface area of 33.5 m2 /g [24]; however, there are some materials with areas much higher than 33.5 m2 /g. Particle size of the adsorbents investigated was 0.5–1.0 mm. 2.2. Potentiometric titration The cation and anion exchange capacity of the investigated sorbents (with respect to H+ and OH− ions correspondingly) was studied by potentiometric titration as described [25,26]. Typically, 1 g of ion exchanger was introduced into a flask containing 50 cm3 of 0.1 M NaCl; then a definite volume of 0.1 M NaOH or HCl was added, the flask was placed in a temperature-controlled orbital shaker for 48 h, and equilibrium pH was measured. Ion exchange capacity, q (mEq/g) was calculated using the equation C(V1 − V2 ) , 1000 × m
2. Materials and methods
q=
2.1. Synthesis of the sorbent and its texture characteristics
where C (mol/ml) is the concentration of the acid/alkali, V1 (ml) is the volume of added NaOH/HCl in the experiment at defined pH, V2 (ml) is the volume of NaOH/HCl added in the blank experiment at the same pH, and m (mg) is the mass of the sorbent.
Chloride salts of iron and aluminum (FeCl3 ·6H2 O and AlCl3 ·6H2 O) and a water solution of ammonia (25%) were used to synthesize the novel ion exchangers. All reagents
(1)
N.I. Chubar et al. / Journal of Colloid and Interface Science 291 (2005) 67–74
2.3. ζ -Potential studies ζ -Potential measurements were carried out with a Microelectrophoretic zeta potential analyzer (Mark II produced by Rank Brothers Co.). The samples were previously prepared as follows: Two hundred milligrams of the sorbent was added to the flask containing 100 ml of water, the pH of the solution was adjusted with 0.1 M HCl or NaOH solution, and the flask was placed in a temperature-controlled orbital shaker for 48 h at 22 ± 2 ◦ C. The electrophoretic mobility of the dispersion transferred into the microelectrophoretric cell was measured. The ζ potential was calculated using the Smoluchovski equation [27,28]. 2.4. Batch sorption experiment Experiments were carried out in Erlenmeyer flasks placed in a temperature-controlled orbital shaker (stirring speed of 150 min−1 ) for 48 h as it had been established previously that 48 h is equilibrium time. Two hundred milligrams of the sorbent was equilibrated to 100 ml of the anion solution at 22 ± 2 ◦ C using the background electrolyte 0.01 M NaNO3 . It was shown separately that NO− 3 is not adsorbed by the surface of this ion exchanger and cannot compete with halides for sorption sites. A suitable volume of acid (HNO3 ) or base (NaOH) solution was added to adjust pH, which was measured with a digital pH meter (TOA Electronics Ltd., HM-35 V). The adsorbent was removed by centrifugation, and the concentration of the corresponding anion remaining in the supernatant was determined by single-column ion chromatography with a low-conductivity mobile phase. The latter consisted of a mixture of 4 mM p-hydroxybenzoic acid and 0.5 mM sodium benzoate (pH 9.0 adjusted with 1 M NaOH) delivered at the flow rate of 2.3 ml/min. The ion chromatograph consisted of an SSI 222D Pump (SSI, State Colleague, PA, USA) and a Wescan 315 conductivity detector (Alltech, Deerfield, IL, USA) maintained at 25 ± 0.1 ◦ C. A Rheodyne 9125 six-port highpressure switching injection valve (Rheodyne, Cotati, CA, USA) was assembled with a 50-µl injection loop. The separation column Hamilton PRP-X100, 150 × 4.1 mm (Hamilton, Reno, NV, USA), was packed with spherical 10-µm polystyrene–divinylbenzene trimethylbenzene trimethylammonium anion exchanger (Hamilton, Reno, NV, USA). The capacity of the column was 0.19 ± 0.02 mEq/g. Data acquisition was performed using a Hewlett–Packard integrator (Model HP 3396 II, Hewlett–Packard, Avondale, PA, USA). The final pH of the solutions was also measured. The experiments were repeated two or three times and the average values were used for data treatment. Anion uptake was determined according to the equation (C0 − Ceq )V , (2) M where q (mg/g) is the amount of anion adsorbed per gram of oxide hydrated, C0 (mg/ml) is the initial concentration of q=
69
the anion in solution, Ceq (mg/ml) is the final (or equilibrium) concentration of the anion in solution, V (ml) is the volume of the solution, and M (g) is the mass of the sorbent.
2.5. Fitting the isotherms to the Langmuir model To describe the mechanism of sorption and to draw up the affinity constants, the experimental results were fit to the linear form by the Langmuir equation, Ceq Ceq 1 = + , qeq Kqmax qmax
(3)
where Ceq (mg/dm3 ) is the equilibrium concentration of the anion in solution, qeq (mg/g) is the correspondent capacity of the sorbent, qmax (mg/g) is the maximum anion concentration sorbed by the ion exchanger, and K (dm3 mg−1 ) is the affinity constant [27,28]. 2.6. Kinetics of the sorption of the investigated anion The kinetics of fluoride and bromide ion sorption was fitted well to a pseudo-second-order mechanism corresponding to the equation dy (4) = Kc (1 − y)2 , dt where Kc is the kinetic constant. Solving this ordinary differential equation alone with the conditions that t = 0, y = 0, Eq. (6) can be linearized into the form, 1 t 1 = + t, q Kc qe qe
(5)
which allows calculation of qe and Kc , respectively, from the slope and intercept of the straight line. Ho [29] was the first to develop the pseudo-second-order kinetic expression for sorption systems, in which chemical sorption is the rate-limiting step; since its development this model has been applied in a number of studies [30–32] including those for sorption of anions [32]. The pseudo-first-order model developed by Lagergren (1998) did not work as well in our experimental data. That is why we described the data with a pseudo-second-order mechanism.
3. Results and discussion 3.1. Surface studies of Fe2 O3 ·Al2 O3 ·xH2 O It is known that hydrated metal oxides are ampholites with cation and anion exchange properties. That is why the materials investigated have shown both cation and anion exchange capacity, which reached values of 1.38 and 1.8 mEq/g, respectively (Fig. 1a). The isoelectric point of the sorbents was at neutral pH.
70
N.I. Chubar et al. / Journal of Colloid and Interface Science 291 (2005) 67–74
(a)
Fig. 2. Plot of sorption of F− , Br− , Cl− , and BrO− 3 versus pH on Fe2 O3 · Al2 O3 ·xH2 O. Conditions of the experiment: 48-h shaking time, 20 ± 2 ◦ C, 120 mg/L initial concentration of each ion, 0.01 M NaNO3 background electrolyte, and 200 mg/L solid concentration.
(b) Fig. 1. Potentiometric titration curve (a) and effect of pH on ζ potential of Fe2 O3 ·Al2 O3 ·xH2 O at background electrolyte 0.01 M NaNO3 (b).
The effect of pH on ζ potential on hydrous oxides of Fe and Al is illustrated in Fig. 1b. The point of zero charge was at pH 3.6. The surface of the materials was negatively charged at pH higher than 3.6. The maximum negative charge was at pH 9. 3.2. Influence of pH on anion sorption Fig. 2 shows the pH effect of anion sorption on Fe2 O3 · Al2 O3 ·xH2 O (mentioned as Fe–Al); the initial concentration of all ions was 120 mg/L. It was found that the sorption capacity of F− on the double hydrous oxide investigated was higher compared with that of the other halides. The higher affinity of the sorbent to fluoride is due to possible formation of insoluble iron fluoride and less soluble aluminum fluorides. The technologically important (for drinking water treatment) pH effect on fluoride, bromide, and bromate sorption on the adsorbent investigated was established. Typically, sorption of anions on individual hydrous oxides is very dependent on pH. For example, F− sorption on the natural mineral goethite [32] and acid-treated spent bleaching earth [33] takes place in acidic solutions only. Sorption of F− decreases slowly with increasing pH, and the anions are not adsorbed at all at pH 7. Fe–Al could take up F− (as well as Br− and BrO− 3 ) (Fig. 2) in the pH range 3–8.5. In acidic solutions at pH 3–4, the sorption capacity of the adsorbent investigated toward fluoride was twice as high as in neutral and alkaline solutions. At pH 4–5 a sharp decrease in sorption from 70 to 37 mg/g takes place. At pH 7.5–8 some decrease in the sorption capacity of the sorbent (near 20 mg/g) occurs. pH < 5 is the most favorable for fluoride uptake due to the lower concentration of OH− ions, which
compete with fluoride for sorption sites with increasing concentration of OH− in the solution and reduce the capacity for sorption of F− . Speciation of the investigated anion is the second reason for higher fluoride sorption at pH < 5 when it exists as HF. Maximal sorption capacity is at pH 4, at which this is the predominant form of the anion. The adsorption of F− can be described as an exchange reaction against OH of the surface groups. IR analysis by Hiemstra and colleagues [32], who investigated sorption of F− on goethite, confirmed that the main reaction of such an adsorption process is singly coordinated FeOH groups. They formulated the formation of FeF−1/2 as FeOH−1/2 + H+ (aq) + F− (aq) FeF−1/2 + H2 O.
(6)
Fluoride species at pH < 5 (HF) are the easiest adsorbing due to formation of water molecules in accordance with the reaction [1]. Sorption methods are not widely accepted for removal of Cl− . We tried to assess the sorption of this anion on Fe–Al to compare the data with data for other halides. Fig. 2 shows that sorption of Cl− on Fe–Al depends very much on pH, and this anion can be adsorbed at pH 3–4 followed by sharp degrees of sorption that become zero at pH 5. This can be explained: ion exchange is the main mechanism for chloride adsorption on the materials investigated and Cl− cannot compete with increasing OH− concentration at higher pH. The pH effect of Br− sorption was similar to the pH effect of F− sorption. Maximum bromide ion sorption also occurred at pH < 5 due to the lower concentration of OH− ions, which compete for sorption sites, and the predominant existence of fluoride as HBr species. Sorption capacity then decreased slowly and reached a small, second peak at pH 9. Sorption of bromate ions was approximately equal through the whole pH interval investigated (3–10), with some small maxima at pH 3 and 9–10. This anion has another mechanism of sorption (confirmed by isotherms, see Fig. 8) which should be studied separately.
N.I. Chubar et al. / Journal of Colloid and Interface Science 291 (2005) 67–74
71
Fig. 3. Scheme of the model cluster of 4Fe–4Al.
3.3. Modeling of the sorption mechanism using quantum-chemistry program The ability of the sorbent to take up F− , Br− , and BrO− 3 ions at pH > 5 can be explained by the fact that pH complexation plays an important role in the binding of these anions to the surface. Most probably, the anions formed complexes with Al atoms due to the higher stability of AlF2+ , AlF+ 2, and AlF3 (pK 6.13, 5.02, and 3.83, respectively) as compared with the stability of FeF2+ , FeF+ 2 , and FeF3 (pK 5.17, 3.83, and 2.91, respectively) [34]. Additional evidence of this suggestion was drawn from some modeling of the adsorbent (4Fe–4Al) structure with HyperChem7 (Figs. 3 and 4). Fig. 3 shows the model cluster of the synthesized sorbent containing equal numbers of iron and aluminum atoms. The electronic structure of this cluster (geometry and charge distribution) has been calculated using the PM3 method. In this model cluster, an atom of aluminum is 3-coordinated, whereas an atom of iron has octahedral and tetrahedral environments. Oxygen is 2- and 3-coordinated here. The charges on atoms of aluminum, iron, and also oxygen of hydroxyl groups linked with aluminum and iron have been shown. There is a large positive charge (near +e) on the aluminum atom, but a small negative charge on the iron atoms. This proves that electronic density has been transfered from oxygen atoms linked with iron. Analysis of the local functions (on atoms) of the density of states (DOS) has shown that the atoms of oxygen and iron have considerable electron-donor properties, whereas aluminum and oxygen have electron-acceptor properties. It is noteworthy that octahedral atoms of iron have more electronacceptor properties but tetrahedral iron has more electrondonor properties. Thus , 3-coordinated atoms of oxygen have the highest electron-donor and electron-acceptor properties. Oxygen atoms of hydroxyl groups are not responsible for electron-acceptor properties.
Fig. 4. Local functions of the density of states (DOS) of cluster 4Fe–4Al.
The charge distribution on the surface, shown in Fig. 1, indicates that considerable positive charges, concentrated on aluminum, are the most probable (reasonable) coordination sites for negatively charged halides. The mechanism of this reaction has been demonstrated by interaction of NaF with the model cluster investigated (Fig. 3). Taking into account that dipole moments of the cluster and NaF are correspondingly 7.07 and 8.11 D, electrostatic interactions, which cause the molecule NaF to orientate to and approach the atom of Al having the biggest charge (q = 1.117e) on the cluster, will dominate at considerable distances between NaF and the cluster (1, 2, and 3 in Fig. 5). Optimization of the geometry of the cluster (3, Fig. 5) by the PM3 method allows us to conclude that during interaction, a transition reactive complex with 4-coordinated Al (AlO3 F) is formed. Such an interaction leads to redistribution of the charge density, separation of Na+ ions, and formation of hydrated cations of Na+ , which interact with OH− of 4-coordinated Al to form NaOH (5 and 6 in Fig. 5). Isolation of OH− and alkalization of the experimental solution are also confirmed by the experimental data. A final result of this interaction is a cluster with bound fluoride. It is worth noting that gain in heat formation of the F-containing cluster as compared with the initial cluster is 37 kcal/mol. Formation of clusters with bound Cl and Br leads to decrease in heat formation (49 and 36 kcal/mol). This means that the binding of fluoride to the surface of the cluster investigated occurs much more easily than the binding of Br and Cl, which is in good correlation with the experimental data in this article.
72
N.I. Chubar et al. / Journal of Colloid and Interface Science 291 (2005) 67–74
Fig. 5. Mechanism of fluoride binding to the surface of 4Fe–4Al.
3.4. Sorption kinetics Fluoride and bromide sorption was found to be relatively fast kinetically. All fluoride ions present in the solution (200 mg/L) with buffer background electrolyte were adsorbed for 20 min. In the first 10 min 50% of the bromide ions and 75% of the fluoride ions were adsorbed (Fig. 6). The kinetics data fit well to the pseudo-second-order model. The rate coefficients of fluoride and bromide sorption calculated from the model were 0.40 and 0.16 min−1 , respectively. Sorption of F− was 2.5 times faster than sorption of Br− under the same conditions. For comparison, adsorption of F− on the ion exchanger investigated was slower than on new ion exchange fiber [15] but slightly faster than on treated spent bleaching earth [33].
(a)
3.5. Isotherms The isotherm of fluoride sorption (Fig. 7), which is very close to the y axis as compared with the curves obtained for the other anions investigated in this article, is evidence of the high affinity of the adsorbent investigated to F− . The capacity of the ion exchanger for sorption of F− (4.2 mmol/g) was four times higher as compared with the sorption capacity of the new ion exchange fiber [15] and 10 times higher than that of goethite [32]. Hiemstra and colleagues [32] studied the mechanism of F− sorption using the mathematical modeling approach CD-MUSIC. We suppose that similar processes occur on Fe–Al. It was concluded that at the lower concentration of F− in the solution, these ions were exchanged with OH− , coordinated on the surface, in a one-charge scheme.
(b) Fig. 6. Kinetics of F− and Br− sorption (a) on Fe2 O3 ·Al2 O3 ·xH2 O and fitting of the sorption kinetics to the pseudo-second-order model (b). Conditions of the experiment: initial concentration 200 mg/L; 0.01 M NaNO3 background electrolyte, pH 4; 20 ± 2 ◦ C.
At higher concentrations of fluoride ions, OH− ions, coordinated on the two-point scheme, take part in the ion exchange with F− . Redistribution of the surface charge takes place
N.I. Chubar et al. / Journal of Colloid and Interface Science 291 (2005) 67–74
(a)
73
− Fig. 8. Isotherms of Br− , BrO− 3 , and Cl sorption on Fe2 O3 ·Al2 O3 ·xH2 O. Conditions of the experiment: 0.01 M NaNO3 background electrolyte, pH 4, 48-h shaking time; 20 ± 2 ◦ C.
(b) Fig. 7. Isotherms of F− sorption on Fe2 O3 ·Al2 O3 ·xH2 O and fitting of the data of curve (a) to the Langmuir model (b). Background electrolyte 0.01 M NaNO3 , pH 4.
on the formation of surface complexes of fluoride. This is explained by a change from donor-type complexes (FeOH, Fe(OH)2 ) to proton-acceptor complexes (FeF3 ). Distribution of the charge in this model is approximately equivalent to the change from donor H bond to acceptor H bond. At very high concentrations of fluoride ions, precipitation of F− into FeF3 occurred. The high concentrations of background electrolytes catalyze this process. Isotherms of F− sorption fit well the Langmuir model with a great correlation coefficient of 0.9987. Maximum sorption capacity calculated from the model was close to the experimental data (90 mg/g), and the affinity constant was 0.52 L/mg, which is 10 times higher than analogous parameters measured for acid-treated spent bleaching earth [33]. Fig. 8 shows the sorption isotherms of Br− , Cl− , and BrO− 3 . The character of these curves is evidence of the lower affinity of the sorbents to these anions, as compared with the close-to-axis y isotherm of F− sorption. Sorption isotherms of BrO− 3 have two plateaus at the equilibrium concentrations of the ions: 20–60 and >90 mg/L. Sorption capacity was 15 and 25 mg/g. Growth of sorption was noted at equilibrium concentrations from 60 to 80 mg/L; then, the second plateaus were found. The difference in the form of isotherms of F− and Br− , on one side, and BrO− 3 , on the other side, can be explained by their different affinities to the anion, different mechanisms of sorption, and speciation. The first plateaus on the curve meant that the most available sorption
Fig. 9. Isotherms of competitive adsorption of Br− and BrO− 3 onto Fe2 O3 · Al2 O3 ·xH2 O under the conditions of their combined presence in the solution at the same concentrations (0.01 M NaNO3 background electrolyte, 48-h shaking time).
sites had been saturated. At the second plateaus, adsorption was taking place on the less favorable sites. Adsorption of bromide was higher as compared with adsorption of bromate ions as confirmed by the study of the competitive adsorption of these two ions (Fig. 9). Investigation of the competitive adsorption of Br− and BrO− 3 added in equal concentrations to experimental solutions has shown that at lower concentrations of the anions investigated (Ceq < 40 mg/L), bromate dominated in the competitive adsorption with bromide ions. At higher concentrations of the anions (>70 mg/L), on the contrary, sorption of BrO− 3 was already zero and Br− was the only ion adsorbed to the surface of the ion exchanger investigated. Such data on competitive adsorption are in good agreement with isotherms for each anion adsorption shown in Fig. 8.
4. Conclusions The anion exchange capacity of the novel inorganic ionite based on the mixed oxide hydrate of iron and aluminum (equivalent ratio) was 1.8 mEq/g, in accordance with the potentiometric titration data. The pH effect of F− and Br− was dependent on ion speciation; however, sufficient capacity for sorption of these anions was observed in the pH
74
N.I. Chubar et al. / Journal of Colloid and Interface Science 291 (2005) 67–74
range 3–10. Adsorption of fluoride (88 mg/g) was the highest among the anions investigated and fit well the Langmuir model with a high affinity constant. The kinetics of F− and Br− sorption was subject to the pseudo-second-order model with rate coefficients of 0.40 and 0.16 min−1 , respectively. Isotherms for the competitive adsorption of Br− and BrO− 3 were in the agreement with isotherms for each ion: at lower concentrations of the ions in the solution (<40 mg/L), bromate dominated; then its sorption was depressed completely by bromide ions. The model cluster of 4Fe–4Al was optimized with HyperChem7 (PM3 method) and the charge distribution on the cluster was obtained. Analysis of the local (on atoms) functions of the state density has shown that 3-coordinated oxygen atoms have the highest electron-donor and electron-acceptor properties. Modeling of the NaF interaction with the model cluster surface has shown that binding of F takes place on the aluminum atoms and this interaction is energetically favorable (gain of heat formation) for fluoride but not for bromide and chloride ion binding.
Acknowledgments Work was supported by a NATO grant to Dr. N. Chubar, Aristotle University, Thessaloniki (Greece), by STCU Collaborative Grant N1739 (manager: Professor V.V. Strelko), and by Bilateral Collaborative Grant M/275 (manager: N. Chubar).
References [1] B.S. Zhang, Research on Endemic Fluoride Toxicosis and Industrial Fluoride Pollution, Chinese Environmental Sciences Press, Beijing, 1994. [2] D. Ró˙zycka, M. Grobelny, Chemik 4 (1998) 92–93. [3] S.K. Sahu, S.S. Pati, R.K. Badapanda, Environ. Ecol. 16 (1998) 169– 171. [4] A.K.M. Arnesen, Water Air Soil Pollut. 103 (1998) 375–388. [5] Y. Kurokawa, A. Maekawa, M. Takahashi, Y. Kayashi, Environ. Health Perspect. 87 (1990) 309–335. [6] R.L. Melnick, Regul. Toxic. Pharmacol. 16 (1992) 111–125.
[7] J. Lawrence, H. Tosine, F.I. Onuska, M.E. Comba, Sci. Eng. 2 (1980) 55–64. [8] A. Mills, A. Belgrhazi, D. Rodman, Water Res. 30 (9) (1996) 1973– 1978. [9] J.G. Jacangelo, J. Am. Water Works Assoc. 87 (1995) 64–77. [10] P. Calza, E. Pelizzetti, Pure Appl. Chem. 73 (2001) 1839–1848. [11] J.H. Harmon, S.G. Kalichman, J. Am. Water Works Assoc. 57 (1965) 245. [12] E. Bellack, J. Am. Water Works Assoc. 63 (1971) 454. [13] D.S. Bhargava, S.B. Sheldarkar, Environ. Pollut. 76 (1992) 51. [14] M. Siddiqui, W. Zhai, G. Amy, C. Mysore, Water Res. 30 (1996) 1651–1660. [15] L. Ruixia, J. Guo, T.J. Hongxiao, J. Colloid Interface Sci. 248 (2002) 268. [16] J.M. DeMarco, A.K. SenGupta, J.E. Greenleaf, Water Res. 37 (2003) 164. [17] N. Gangoli, G. Thodos, J. Water Pollut. Control Fed. 45 (1973) 842. [18] S. Tokunaga, S.A. Wasay, S.W. Park, Water Sci. Technol. 35 (1997) 71. [19] S.A. Wasay, A. Tokunaga, S.W. Park, Sep. Sci. Technol. 31 (1996) 1501. [20] I.Z. Zhuravlev, V.A. Kanibolotsky, V.V. Strelko, A.I. Bortun, L.I. Bortun, S.A. Khainokov, A. Clearfield, J. Solv. Extract. Ion Exchange 17 (1999) 635. [21] A.I. Bortun, S.A. Khainokov, V.V. Strelko, I.A. Farbun, in: J.A. Grieg (Ed.), Proc. Ion Exchange Developments and Application, Cambridge, UK, 1996, pp. 305–310. [22] S. Brunauer, P.H. Emmeet, E. Teller, J. Am. Chem. Soc. 60 (1938) 309. [23] N.B. Keltsev, The Basis of Sorption Techniques, Chemistry, Moscow, 1984. [24] G. Yan, A. Mucci, Chem. Geol. 199 (2003) 91. [25] F. Helfrich, Ionites, Foreign Literature, Moscow, 1962. [26] R. Grissbach, Theory and Practice of the Ion Exchange, Foreign Literature, Moscow, 1963. [27] L. Svarovsky, Solid–Liquid Separation, third ed., Butterworths, London, 1990. [28] S.S. Voyutsky, Course of Colloidal Chemistry, Chemistry, Moscow, 1964. [29] Y.S. Ho, Ph.D. thesis, University of Birmingham, UK, 1995. [30] Y.S. Ho, G. McKay, J. Water Res. 34 (2000) 735. [31] R. Machado, J.M.R. Carvalho, M.J.N. Correia, J. Chem. Technol. Biotechnol. 77 (2002) 1340. [32] T. Hiemstra, W.H. Van Riemsdijk, J. Colloid Interface Sci. 225 (2000) 94–104. [33] M. Mahramanlioglu, I. Kizilcikli, I.O. Bicer, J. Fluoride Chem. 115 (2002) 41–47. [34] I.T. Goronovsky, Y.P. Nazarenko, E.F. Nekryach, in: O.D. Kurilenko (Ed.), Short Reference Book on Chemistry, Naukova Dumka, Kiev, 1974, p. 990.