Adsorption of anionic dyes onto natural, thermally and chemically modified smectite clays

Abstract The aim of this study was to determine the adsorption capacity of the smectite clays (from the overburden of the lignite deposit in Belchatow) for two anionic dyes, i.e. Reactive Blue 81 (RB-81) and Direct Blue 74 (DB-74). Additionally, the influence of the thermal and chemical (acid and alkali) clay modifications on the amount of bonded dyes was investigated. The adsorption capacity of the clay (natural and modified) was different for studied dyes and depended on the initial concentration and modification type. All the modified clays adsorbed the dyes at pH>pHPZC as the negatively charged surfaces of their particles (in accordance with the formula: AOH ↔ AO- + H+) prevented the formation of electrostatic bonds between the anionic dyes and the clay surface. The dyes were mainly bound with the hydrogen bonds forming between the donor groups in the dyes and the acceptor groups (-SiO and -Al2OH) in the clays. The coefficients in the adsorption isotherms were estimated with the linear and non-linear regression. The linear regression method was found that the Freundlich and Dubinin-Radushkevich isotherms described the dye sorption much better than the Langmuir model. On the other hand, all three models described well the experimental data in the non-linear regression method. Furthermore, the 1/n value (<1) obtained from the Freundlich equation for all the dye-sorbent systems indicated the favorable sorption.


INTRODUCTION
Dyes are one of the organic contaminants found in wastewater. They are generated by the paint, textile, cosmetic, paper and food industries 1-3 . At present, over 100.000 colors are used for commercial purposes. Their annual production is estimated at more than 7 x 10 5 Mg. 10-15% of dyes used in the coloring processes penetrate the water environment. They pose a threat because even low amounts of dyes cause water coloration and hinder the photosynthesis processes taking place in the aquatic organisms 4 . Dyes can also have a negative impact on the human health. They can cause skin and eye irritation. Some of them are toxic and carcinogenic 5-7 . In the textile industry, reactive, direct, acidic and alkaline dyes (whose character is cationic or anionic) are used to dye cellulose, protein and polyamide fi bers.
The wastewater containing dyes should be treated before its discharge into water or sewerage systems. The adsorption techniques based on active carbons make effective methods of dye removal from water and wastewater at low and high concentrations 8 . However, the high cost of the activated carbon production and problems with its regeneration after the sorption capacity exhaustion make it necessary to look for more effective, easily accessible and economical sorption materials. Such materials can come both from natural and waste sources and need only simple preliminary preparation. These include zeolites, bentonite and bentonite clays, and smectite clays 5, 6, 9, 10-15 . The smectite family minerals build clay rocks. They have high specifi c surface area and cation exchange capacity. Their layer structure indicates the possibility to increase these parameters due to thermal or chemical modifi cations.
Many studies demonstrate that smectite clays have a high sorption capacity for basic dyes 15-17 . However, there are very few works on the adsorption of anionic dyes. Large amounts of the anionic dyes are bound by the smectite clays at pH<2 18 . The sorption capacity also depends on the mineral composition of the clay, including the kaolinite content, i.e. Reactive Blue 81 and Direct Blue 74. It was also conducted to determine the effects of thermal and chemical modifi cations of the clay on its sorption capacity. The estimated sorption parameters helped to determine the bonding mechanism in the investigated dyes.

Clay sorbents
The smectite clay from the Belchatow deposit of the lignite mine "Belchatow" was used in the study. It was dried at room temperature (25 ± 2°C) and crushed to a fraction (<0.5 mm). The investigations also embraced the clay forms after modifi cations. They included the thermal modifi cation (heating at 250°C); the acidic one (with 16% H 2 SO 4 solution at 96°C for 8 hours, the ratio of the solid phase to solution = 1:5) 21 ; and alkaline one (with 5M NaOH solution at 90°C, the ratio of the solid phase to solution = 1:5) 22 .
Typical methods applied in the mineralogical research were used to determine the characteristics of the natural and modifi ed clay samples 23 . These encompassed mineral and chemical compositions, major functional groups and physicochemical properties (total (SSA H 2 O) and external (SSA N 2 ) specifi c surface area, porosity (n), an average pore diameter (d), cation exchange capacity (CEC) and the exchangeable cations). Moreover, pH values and the isoelectric point (pH PZC ), defi ned by the pH in which the charge of the colloidal particles is equal to zero, were also determined (Table 1).
The structural formulas of the tested dyes are shown in Figure 1.

Experimental method
The sorption capacities of the natural and modifi ed clays for the RB-81 and DB-74 dyes were determined with the batch method under static conditions contacting the solid phase : the dye solution. Aqueous dye solutions (initial concentrations of 1, 10, 25, 50, 100, 150, 250, 500, 750, and 1000 mg · dm -3 ) obtained though diluting the stock solution with distilled water (1000 mg · dm -3 ). The solid phase-solution ratio was 1:20 (i.e. 1 g of the clay sample + 20 ml of dye solution at the adequate concentration level). The shaking time was 24 hours. Then, the solution was separated from the solid phase during 20-minute centrifugation (4000 rpm).
The concentrations of the RB-81 and DB-74 dyes in the initial (C 0 ) and equilibrium (C eq ) solutions were determined with the UV-Vis spectrometry (Varian Cary 50 Scan UV-VIS spectrometer) at wavelengths of 584 nm and 582 nm, respectively. Moreover, the pH values were measured in all the equilibrium solutions.
The dye amount (q) adsorbed onto the clay phase and the percentage removal of dye from solution (R) were calculated from the following formulas: (2) where: q -the amount of the dye adsorbed by the clay particles (mg . kg -1 ), R -percentage removal of the dye from the solution (%), C 0 -the initial concentration of the dye solution (mg . dm -3 ), C eq -the equilibrium concentration of the dye solution (mg . dm -3 ), The obtained results allowed the determination of the sorption isotherms in the system of q = f(C eq ) and dye removal curves in the system of R = f(C 0 ).

Estimating the sorption parameters
To interpret the experimental data on the sorption of dyes onto clay particles, three non-linear isotherms 25-27 were used: -the Freundlich adsorption isotherm: -the Langmuir adsorption isotherm: -the Dubinin-Radushkevich adsorption isotherm: q = q D . exp(-βε 2 ) (5) The parameters in the equations (3-5) were estimated with the linear and non-linear regression methods.
The linear regression method was based on the transformation of the equation to the linear form of the following formulas: -the Freundlich isotherm: where: K F -the sorption capacity coeffi cient (dm 3 . kg -1 ), 1/n -the heterogeneity parameter, the closer the 1/n value to zero is, the more heterogeneous 28 the surface becomes. Moreover, the 1/n value determines the adsorption intensity.
-the Langmuir isotherm: where: q L -the maximum adsorption capacity of the sorbent for a dye (mg . kg -1 ), K L -the adsorption equilibrium constant (dm 3 . mg -1 ).
The values of the 1/n and K F parameters in Eq. (6) were estimated by plotting log(q) vs log(C eq ) (K F = exp(intercept), 1/n = slope). The values of q L and K L in Eq. (7) were calculated from the dependence vs C eq (q L = 1/slope, K L = 1/(intercept x q L ). Finally, the parameters q D and β in Eq. (8) were computed from the correlation: ln(q) vs ε 2 (q D = exp(intercept), β = -(slope).
The computer software Statistica (ver. 9.0) was used to estimate the values of coeffi cients in the sorption equations with the non-linear regression method. It is based on the method of least squares method with the Gauss-Newton estimation algorithm.
The value of the adsorption equilibrium constant K L estimated from Eq. (7), allowed the calculation of the separation factor R L from the equation: (10) where: C 0 -the highest initial dye concentration in the experimental solution (mg . dm -3 ).
The value of β constant was estimated from Eq. (8). It was used to determine free energy E per dye molecule from the equation: Using the estimated values of the 1/n, R L and E parameters, the researchers attempted to determine the mechanism of the dye bonding onto the clay particles and adsorption conditions.
The value of the 1/n parameter <1 (Eq. 6) indicated a favorable sorption process 28, 29 . The value of the R L (Eq. 10) parameter indicated the following adsorption conditions: when R L >1 -adsorption conditions were unfavorable; R L =1 -linear adsorption, 0<R L <1 -adsorption conditions were favorable; and when R L = 0 -irreversible adsorption. The value of the free energy per one dye molecule (E) estimated from Eq. (11) allows differentiation of the physical adsorption (E<8 kJ . mol -1 ) from the ion exchange (8<E<16 kJ . mol -1 ). However, the q L parameter estimated from the Eq. (7) indicated the monolayer capacity, i.e. the maximum sorption capacity of clay for the dye.
The groups were responsible for the surface charge of the particles. They could act as both donors and acceptors of protons in a hydrogen bond.
Moreover, the doublet at 802 and 800 cm -1 and 1448 cm -1 bands pointed to the presence of quartz and calcite admixtures, respectively.
The cation exchange capacity (CEC) of the clay was 82.39 cmol + . kg -1 . The main exchangeable cations were Ca 2+ and Mg 2+ 11 . The external and total specifi c surface areas were 41.38 m 2 . g -1 and 141.83 m 2 . g -1 , respectively. The porosity was 0.0682.
During the clay drying at 250°C, two phenomena occurred, i.e. the dehydration (removal of water molecules) of the smectite mineral and decrease in the peak intensity at 3430 cm -1 and 1654 cm -1 bands, which caused the increase in the porosity value (0.1113) and the pore diameter (0.0322 μm). There was also a nearly two-fold increase in the external surface area (Table 1). The clay particles were treated with a 16% solution of H 2 SO 4 . As a result, the octahedral smectite sheets were destroyed, which was confi rmed by the complete disappearance of the peak intensity of the characteristic band at 873 cm -1 and the decrease in the intensity at 921 cm -1 and 536 cm -1 . There was also a decrease in the vibration intensity at 1054 cm -1 and 474 cm -1 related to the degradation of the tetrahedral sheets. Calcite was dissolved (according to the reaction: CaCO 3 + H 2 SO 4 → Ca 2+ + SO 2+ 4 + CO 2 + H 2 O), which was demonstrated by the disappearance of the intensity of a band at 1448 cm -1 . The exchange of the used H 2 SO 4 during the clay modifi cation was performed in accordance with the methodology and prevented the CaSO 4 precipitation 21 . Due to the octahedral sheets destruction, the percentage of the silica in the samples rose and the value of The variation in the CEC value correlated with the changes in the total specifi c surface area of the samples 30 .

The sorption of the reactive RB-81 and direct DB-74 dyes onto the natural and modifi ed clay particles
The experimental adsorption isotherms of the RB-81 and DB-74 dyes onto smectite clays (natural, thermally and chemically modifi ed clays) are presented in the system of q = f(C eq ) (Fig. 3). The dye removal curves from the solution are given in the system of R = f(C 0 ) (Fig. 4). The results of the pH measurements in the equilibrium solutions after sorption are shown in Figure 5.
When the adsorbed amount of the studied dyes was analyzed (depending on their initial concentrations in the solution), it was found that the clay with the natural sorption complex adsorbed 18.8-2870 mg . kg -1 of the RB-81 dye. This indicated the dye removal from the solution at the level of 94-14.15%. On the other hand, the sorption capacity of the clay for the DB-74 dye was larger and ranged between 19.2 mg . kg -1 and 4700 mg · kg -1 . The dye percentage removal from the solution was between 96 and 22.66%.
Even though the RB-81 pH was acidic (4.97) and the DB-74 was alkaline (8.03), the correlation curves of the pH values and the dye concentration in the equilibrium solution indicated that the sorption process of the te-the calculated SiO 2 /Al 2 O 3 ratio increased to 17.25. The acid modifi cation of clay induced a six-fold porosity growth (0.4268), whereas the average pore diameter increased to 2.1068 μm. Moreover, the external specifi c surface area rose twice (77.44 m 2 · g -1 ). The saturation of the sorption complex with the H + ions resulted in the reductions of the pH (3.39) and CEC value (78.55 cmol + · kg -1 ) ( Table 1).
The studied clay was also modifi ed with the 5M NaOH solution. The process brought about the decrease in the intensity of the bands at 1054 cm -1 and 476 cm -1 , which came from the -Si-O group vibrations, and was related to the destruction of the smectite tetrahedral sheets. Additionally, the peak of a band at 1448 cm -1 , derived from the dissolved calcite (according to reaction: CaCO 3 + NaOH → Na 2 CO 3 + CaO + H 2 O), disappeared. As the tetrahedral sheets were destroyed, the SiO 2 /Al 2 O 3 ratio lowered to 0.82.
The alkaline clay modifi cation induced the eight-fold increase in the porosity (0.5334). The average pore diameter rose to 0.6362 μm, whereas the total and external surface areas increased to 214 m 2 · g -1 and 50.09 m 2 · g -1 , respectively. The saturation of the sorption complex with Na + ions and the increase in the CEC value (96.57 cmol + · kg -1 ) were also observed ( Table 1). and alkali (BC NaOH) The fi ndings demonstrate that the studied anionic dyes could not be bound onto mineral surfaces due to the electrostatic interactions. The dye structures, in which the proton donor groups were present (e.g. -OH hydroxyl and -NH and-NH 2 amino groups), lead to a conclusion that the dyes could be bound with smectite clay through a hydrogen bond. In such a case, the dissociated silanol (-Si-O) and aluminol (-Al-OH) groups acted as the acceptors. Furthermore, the interlayer Ca 2+ ions hydrated by water molecules occurred in the smectite structure. The dissociation degree of those water molecules was approx. 10 7 times higher than the liquid water dissociation degree. The discussed water molecules could be the donor of protons in the hydrogen bonds with the dye sulfonate groups acting as the acceptors. In the RB-81 and DB-74 structures, there were three and four sulfonate groups, respectively. The hydrated interlayer Ca 2+ ions could participate in the dye bonding because of the reaction with the sulfonic anion -SO 3 -(according to the formula: Ca(H 2 O) 6 2+ ↔ -SO 3 -) 19 . However, this process was limited due to the large size of the dye particles.
Heating the clay at 250°C brought about the increase in its sorption capacity for both dyes in the whole range of the initial concentrations (Fig. 4). The maximum sorption capacities of the clay for the RB-81 and DB-74 dyes increased by 31.22% (3766 mg . kg -1 ) and 24.42% (5848 mg . kg -1 ), respectively (Fig. 3). The dyes removal depended on their concentrations in the solution. For the RB-81 it was from 95% to 18.56%, and for the DB-74 it was from 98% to 28.2% (Fig. 4). The sorption processes of both dyes onto the thermally modifi ed clay proceeded at the pH values similar to the natural clay pH. The increase in the sorption capacity of the thermally modifi ed clay for both dyes resulted mainly from the increase in its external surface and porosity ( Table 1).
The acid activation of the clay caused a signifi cant increase in the sorption capacity for both dyes in the whole concentration range. When the initial dye concentration was 1000 mg . dm -3 , the amounts of the bound RB-81 and DB-74 dyes were 9860 mg . kg -1 and 15 324 mg . kg -1 , respectively (Fig. 3). This indicates the increase in the sorption capacity of the acid-modifi ed clay by 244% and 226%, respectively. The calculated percentage removal of the RB-81 and DB-74 dyes from the solution was 96-47.87% and 98-73.89%, respectively. Both dyes were bound in an acid medium at the pH of 4.56-3.62. When taking into account the pH PZC value of the acid-modifi ed clay (3.22), the charge of surface particle was negative during the sorption. Moreover, the dyes were mainly bonded by the hydrogen bonds. The acid modifi cation of the clay resulted in the increases in its external surface, porosity and diameter of the pores that facilitated the access of the dye ions to the surface functional groups. On the other hand, the interlayer cations of the natural smectite clay were replaced with protons in the form of hydronium ions (H 3 O + ). Consequently, this caused increases in both the Brønsted-type acid centers and sorption capacities.
Due to the treatment of clay samples with the 5M NaOH solution, the clay sorption capacity increased by 125% (when compared to the natural clay) for the DB-74 dye and reached its maximum level of 10 600 mg . kg -1 . The percentage removal of the DB -74 dye was 94-51.11%. The amounts of the bonded RB-81 dye were lower (17.5-2652 mg . kg -1 ).
The dye sorption processes occurred at the alkaline solution pH (9.75-10.22), which was lower than the clay pH PZC . Due to the alkaline modifi cation, the tetrahedral sheets of the smectite mineral were destructed and the number of the Si-O groups was considerably reduced. What is more, the access of the dyes to the aluminol groups improved, and the sorption complex was saturated with the Na + ions. When compared with the natural clay, the porosity and total and external surface areas increased. The presence of the Na + ions in the interlayer structures (their ability for the water molecule polarization is lower than that of Ca 2+ ions) was one of the reasons for the lower sorption capacity of the alkali-modifi ed clays.
Importantly, the lower adsorption of acid dyes at alkaline pH was due to the presence of the excess hydroxyl ions competing with the dye anions for the adsorption sites 31 .
When summarizing the effects of the thermal and chemical modifi cations of the smectite clay (BC) on its sorption capacities for the RB-81 and DB-74 dyes, it was found that: BC H 2 SO 4 (9860 mg . kg -1 ) > BC 250 (3766 mg . kg -1 ) > BC (2870 mg . kg -1 ) > BC NaOH (2870 mg . kg -1 ); -the amount of DB-74 increased in the order: BC H 2 SO 4 (15 324 mg . kg -1 ) > BC NaOH (10 600 mg . kg -1 ) > BC 250 (5848 mg . kg -1 ) > BC (4700 mg . kg -1 ). The physicochemical properties of the sorbent were not the only important factor determining the amount of the bonded dyes. Aspects such as the dye structures 31 (including the number of the donor groups and their susceptibility to deprotonation under alkaline conditions), the number of the acceptor groups and their location in the benzene rings 32 also played a signifi cant role.
Both studied dyes had the same number of the donor sites, i.e. the RB-81 dye: one -OH group and two -NH groups; the DB-74 dye: one -OH group and one -NH 2 group with two donors (Fig. 1).
The sulfonic groups (three and four in the RB-81 DB-74 dyes, respectively), nitrogen atoms in the azo groups (one in the RB-81; three in the DB-74) and the chlorotriazo group in the RB-81 dye could also act as acceptors. The differences in the dye amounts adsorbed by the tested clays suggest that the sulfonic groups could be important in the sorption of the anionic dyes.
Moreover, the high bonding capacity of the DB-74 dye under acidic conditions can be explained with the pH infl uence on the functional groups 33, 34 . At pH = 3.39, the RB-81 molecule did not change its structure, whereas the -NH 2 present in the DB-74 structure underwent protonation (-NH 3 + ). As a result, more DB-74 molecules bonded with the sorbent surface (Table 2).
Under alkaline conditions (pH = 10.28), the -NH and -OH groups, present in both dye structures, underwent deprotonation (-N -, -O -), which reduced the number of the donor sites and the capacity of the dyes to bond with the clay sorption centers. Importantly, the -NH 2 (with two donor sites) present in the DB-74 dye did not undergo the reaction 33, 34 . Thus, the dye was effectively bonded.
ations. The determination coeffi cient values ranged from 0.9260 to 0.9691. The 1/n parameter value estimated from the Freundlich equation (Eq.6) was < 1, which indicated a favorable sorption process. The values of the E coeffi cient calculated from Eq. (11) indicated that the physiosorption played a signifi cant role in the adsorption process due to the weak hydrogen bond. The Langmuir equation described the RB-81 dye sorption onto natural and modifi ed clay particles to a much lesser degree. The determination coeffi cients were 0.5571-0.7639. The DB-74 dye sorption onto the natural and alkali-modifi ed clays was well described with the Freundlich and Dubinin--Radushkevich equations. This was confi rmed by the R 2 values (0.9152-0.9910). At the same time, these equations described the DB-74 dye sorption onto the clay modifi ed with the sulfuric acid to a lesser extent (R 2 = 0.87). The 1/n constant values (<1) indicated a favorable sorption process while the calculated E coeffi cient values (> 8 kJ/mol) pointed to the ion exchange.
Moreover, the Freundlich constant K F indicated the sorption capacity of the sorbent. The values of K F are given in Table 3. For BC H 2 SO 4 , they were 121.6 dm 3 . kg -1 and 558.5 dm 3 . kg -1 for the RB-81 and DB-74, respectively. The results show that the acid-activated clay had a very strong adsorption capacity for anionic dyes in the solution 31 .
The analyses of the theoretical isotherms took into account the estimated constant values in the equations in relation to the experimental points. It was found that the dye sorption at the low initial concentration range (up to 100 mg/dm 3 ), was best described with the Freundlich and Dubinin-Radushkevich isotherms. For the high concentration range, the Langmuir isotherm turned out to be the most proper (Fig. 6a).
When using the non-linear regression method to estimate the constant values in the sorption isotherms, it turned out that all three isotherms were well fi tted for the experimental points (Fig. 6b). The Q parameter values, obtained from the Langmuir equation, which indicated the maximum clay sorption capacity for both dyes were approx. 50% higher than the values estimated with the linear regression and 60-100% higher than the experimental values. The 1/n parameter values, estimated from Freundlich equation, pointed to a favorable sorption process of the studied dyes onto natural and activated clays. The estimated values of the E coeffi cient indicated that the physical sorption was the main mechanism of the RB-81 dye bonding, whereas the DB-74 dye was bonded with the ion exchange.
The values in parentheses were obtained with the nonlinear regression method.

CONCLUSIONS
The research into the sorption of the anionic dyes (RB-81 and DB-74) onto the natural and thermally and chemically modifi ed smectite clays showed a signifi cant effect of the acid modifi cation on the amount of bonded dyes. The acid-activated smectite had a strong adsorption capacity for the anionic RB-81 and DB-74 dyes in the solution.
Regardless of the clay modifi cation type, the dye sorption occurred at the pH>pH PZC , which indicated

Estimating parameters in the sorption isotherm equations
The values of the Langmuir (q L , K L ), Freundlich (K F , 1/n) and Dubinin-Radushkevich (q D , β) adsorption isotherms parameters, estimated with the linear and non-linear regression methods, and the values of the determination coeffi cients (R 2 ) are given in Table 3. They allowed the creation of the sorption isotherms in the system of q = f(Ceq) (Fig. 6).
The linear regression method made use of the R 2 values. It was found that the RB-81 dye sorption onto natural and modifi ed clay was best described by the Freundlich and Dubinin-Radushkevich isotherm equ-that the particles of the clay-building minerals had the negative surface charge. It was impossible to bond the dyes by means of the electrostatic interaction with the mineral surfaces. The dyes were mainly bonded with the hydrogen bonds, in which the -OH, -NH and -NH 2 groups acted as the proton donors and the -SiO and -Al 2 OH groups played the role of the proton acceptors. The studied dyes could also be bound due to the interactions of the hydrated interlayer Ca 2+ ions, found in the smectite clay, with the dye -SO 3 groups. However, this process was limited due to the dye molecule sizes. The anionic dye structures played an important role in the dye bonding, including aspects such as the number and type of the donor groups and their deprotonation capacity under alkaline conditions; and the number of the acceptor groups.
The linear regression method was used to estimate the sorption constants in the equations. It was found that the Freundlich and Dubinin-Radushkevich isotherms described the dye sorption much better than the Langmuir model. On the other hand, all three models described well the experimental data in the non-linear regression method. Furthermore, the 1/n value (<1) for all the dye-sorbent systems indicated the favorable sorption.

ACKNOWLEDGMENT
The study was conducted under the statute research in the Institute of Environmental Engineering of the Polish Academy of Sciences in the years 2008-2010.