- Email: [email protected]

Applied Clay Science 42 (2008) 137 – 145 www.elsevier.com/locate/clay

Adsorption modelling of textile dyes by sepiolite Sílvia C.R. Santos, Rui A.R. Boaventura ⁎ LSRE — Laboratory of Separation and Reaction Engineering, Departamento de Engenharia Química, Faculdade de Engenharia da Universidade do Porto, Rua Dr. Roberto Frias 4200-465 Porto, Portugal Received 14 June 2007; received in revised form 8 January 2008; accepted 14 January 2008 Available online 26 January 2008

Abstract Adsorption of Astrazon Red (Basic Red 46) and Sirius Blue (Direct Blue 85) in aqueous solution by sepiolite was studied using Response Surface Methodology. A Box–Behnken design was performed to evaluate the effect of initial dye concentration, initial solution pH and temperature on the amount of dye adsorbed at equilibrium. Quadratic models were used to fit the experimental data obtained for both dyes and their adequacy was statistically validated. Linear and quadratic effects of initial dye concentration and linear effect of initial pH were shown to be very significant for Astrazon Red adsorption. Initial pH played the most important effect on the adsorbed amount of Sirius Blue dye but the linear effect of initial dye concentration, its interaction effect with initial pH and the quadratic effect of pH were shown to be also very significant. © 2008 Elsevier B.V. All rights reserved. Keywords: Adsorption; Dyes; Sepiolite; RSM; Box–Behnken design

1. Introduction Textile industries discharge large amounts of coloured wastewaters due to the unfixed dyes on fibres during the colouring and washing steps. The presence of these pollutants in water streams causes problems related to their carcinogenity, toxicity to aquatic life and the easily detected and undesirable aesthetic aspect. Dyeing effluents are very difficult to treat, due to their resistance to biodegradability, stability to light, heat and oxidizing agents (Ozcan et al., 2005; Sun and Yang, 2003). In general, the treatment of dye-containing effluents is being undertaken by adsorption, oxidation–ozonation, biological processes, coagulation–flocculation and membrane processes (Walker and Weatherley, 1999). Adsorption has been proved to be an efficient technique, providing a low-cost maintenance and an easy operation (Bhatnagar and Jain, 2005; Garg et al., 2003). Activated carbon, the most widely used adsorbent (Ozcan and Ozcan, 2005), has been successful (Kadirvelu et al., 2003), but it has the disadvantages of the high costs associated with its replacement and regeneration. This limitation has encouraged

⁎ Corresponding author. Tel.: +351 22 5081683; fax: +351 22 508 1674. E-mail address: [email protected] (R.A.R. Boaventura). 0169-1317/$ - see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.clay.2008.01.002

the search for inexpensive and readily available adsorbents like natural and waste materials. Studies with clays (Ozcan and Ozcan, 2005, 2004; Ozdemir et al., 2004; Wibulswas, 2004), chitin (Figueiredo et al., 2005), algae (Aksu and Tezer, 2000), agricultural waste residues (Robinson et al., 2002), fly ash (Acemioglu, 2004), hydroxide metal sludge (Netpradit et al., 2004), and sewage sludge (Martin et al., 2003), have been found in literature. Sepiolite differs from other 2:1 minerals, because of the discontinuity of the silica sheets, which gives rise to the referred structural tunnels (Bilgic, 2005). Silanol groups (SiOH) are present at the border of each block in the external surface of the silicate (Alrichs et al., 1975) and act as neutral sites for adsorption. It's well known that adsorption is dependent on various factors: adsorbent dosage, initial adsorbate concentration, contact time, temperature, particle size, pH and ionic strength. In conventional methods used to determine the influence of each one of these factors, experiments were carried out varying systematically the studied factor and keeping constant the others. This should be repeated to all the influence factors, resulting in an unreliable number of experiments. In addition, this exhaustive procedure is not able to find combined effect of the factors.

138

S.C.R. Santos, R.A.R. Boaventura / Applied Clay Science 42 (2008) 137–145

Response Surface Methodology, RSM, is an experimental technique that uses mathematical and statistical techniques to analyze the influence of independent variables on a specific dependent variable (response) (Montgomery, 1991). Using RSM, it is possible to estimate linear, interaction and quadratic effects of the factors and a prediction model for the response. In this way, RSM designs could be used to find improved or optimal process settings in an efficient use of the resources. Experimental data required are dependent on the chosen design, Central Composite or Box–Behnken designs (Box and Behnken, 1960). These are different in the number of runs required and in the combinations of the levels used in the experiments. In the case of three studied factors, Box–Behnken design offers some advantages as requires few runs (15 runs). In short, RSM involves the following steps: performing the statistically designed experiments according to the design, factors and levels selected; estimating the coefficients of the mathematical model to predict the response and check its adequacy (Annadurai and Sheeja, 1998). RSM methodology has been applied to model and optimize several wastewater treatment processes including adsorption (Annadurai et al., 2002; Annadurai and Sheeja, 1998; Mohapatra and Gupta, 2005; Ravikumar et al., 2005a,b), Fenton's oxidation (Ahmadi et al., 2005; Catalkaya and Kargi, 2007) and photocatalytic decolourization (Liu and Chiou, 2005). Ravikumar et al. (2005a,b) used a 24 full factorial central composite design to study the combined effect of pH, temperature, particle size and time on basic and acid dyes adsorption by a hybrid adsorbent (mixture of carbon and fly ash). Box–Behnken design was applied to the adsorption of the reactive dye Verofix Red by chitin and a quadratic polynomial equation was achieved as the correlation between the amount of adsorbed dye and pH, temperature and particle size (Annadurai and Sheeja, 1998). RSM was also applied to model Acid Orange 7 dye adsorption removal by spent brewery grains (in dried state and previously acid hydrolysed) as function of adsorbent amount, hydrolysis time, adsorption time, dye concentration and pH (Silva et al., 2004). In this work, adsorption of a basic and a direct dye by sepiolite was studied by RSM, using a Box–Behnken design. 2. Materials and methods 2.1. Dyes Astrazon Red FBL 200% 03 (C. I. Basic Red 46) and Sirius Blue K-CFN (C. I. Direct Blue 85), kindly supplied by Dystar, were selected as adsorbates. The dyes, both from azoic chemical class, were used in aqueous solutions and were analyzed by spectrometry (Helius Alpha Unicam) at the wavelengths corresponding to their maximum absorbance (λmax). In the pH range of 2.0– 9.0, Astrazon Red aqueous solutions presented no dependence between absorbance at 525 nm and pH and a constant absorbance coefficient of 0.0908 cm− 1/(mg/L). Sirius Blue has its maximum absorbance at 590 nm in the pH range of 3.5–10.5 and an absorbance coefficient of 0.0304 cm− 1/(mg/L). For pH values above 9.0 (Astrazon Red) and pH below 3.5 (Sirius Blue), λmax and thus absorbance become very unstable with pH.

2.2. Sepiolite A commercial sepiolite of grain size in the range of 0.250–0.600 mm, obtained from Tolsa (Spain), was used as adsorbent in the dried state. X-Ray Diffraction (DRX) indicates a mineralogical composition of 78% of sepiolite,

9% of K-feldspar, 5% of quartz, 4% of mica and 4% of calcite. Chemical composition, determined by X-Ray Fluorescence, is presented in Table 1. Physical textural properties of sepiolite measured by nitrogen adsorption at 77 K, mercury porosimetry and helium picnometry, presented in Table 2, reveal an essentially mesoporous material. Cation exchange capacity (CEC) was determined to be 0.27 mmol/g (procedure based on EPA Method no. 9081).

2.3. Experimental design A Box–Behnken design of three factors at three levels was chosen. Initial dye concentration (Cin), initial solution pH (pHin) and temperature (T) were selected as independent factors and the amount of dye adsorbed at equilibrium (q) was chosen to be the response of the process (dependent variable). Real values of the factors were selected at three levels (low, medium and high) and were converted into dimensionless variables: the lowest value was coded by (− 1), medium value by 0 and the highest value by 1 (Table 3). Dimensionless variables are then designated here by x1, x2 and x3, respectively for Cin, pHin and T. Initial dye concentration range was selected to be close to the typical concentration values found in textile wastewaters. For Astrazon Red cationic dye, typical concentrations were lower than for direct dye but a minimum of about 65 mg/L should be used to obtain some residual dye at the equilibrium. A wide pH range was selected for both dyes. For temperature levels selection, it was taken into account that 20–40 °C are typical values found in textile dyeing wastewaters. The three-factor Box–Behnken design requires 15 runs (including three center points) that were employed with a replicate and thus a total of 30 experiments were done for the adsorption of each dye. Table 4 presents the coded factors levels used in these experiments.

2.4. Adsorption runs Dye solutions with the desired concentrations were prepared and initial pH adjusted with NaOH or HCl. 100 mL of solution was shaken, at the constant required temperatures, with 100 mg (for Astrazon Red adsorption) or 50 mg (for Sirius Blue) of sepiolite. Adsorbent dosages were chosen in order to guarantee measurable concentrations at equilibrium (thus minimizing the errors associated to the measurement of low dye concentrations or to slight differences between the initial and equilibrium concentrations). Previously performed kinetic adsorption experiments revealed that for Astrazon Red adsorption, a contact time of 15 h was enough to attain equilibrium but a period of 7 days was required for Sirius Blue. After allowing the suspensions to contact up to equilibrium, samples were taken out and centrifuged for 6 min at 13,400 rpm (Mini Spin Eppendorff). The equilibrium dye concentration was measured and the amount of dye adsorbed at equilibrium was then calculated. The values of pH (not adjusted during the contact time) were measured and recorded at equilibrium. For Astrazon Red, when an initial pH value of 2.0 was used, no variation was observed along time. However, for other initial pH values an increase to 8.2–9.2 was recorded, due to the sepiolite strong buffer capacity. For Sirius Blue adsorption, when the initial pH was adjusted to 10.5, practically the same value

Table 1 Chemical analysis of sepiolite by XRF wt.% SiO2 Al2O3 Fe2O3 MnO CaO MgO Na2O K2O TiO2 P2O5 Other constituents Loss on ignition

52 4.8 1.4 0.04 3.2 16 0.42 1.3 0.18 0.09 0.57 20

S.C.R. Santos, R.A.R. Boaventura / Applied Clay Science 42 (2008) 137–145 Table 2 Physical textural properties of sepiolite Nitrogen adsorption Hg porosimetry

He picnometry

Table 4 Box–Behnken design: experimental conditions, results and predicted values

SB.E.T. (m2/g)

107

Vmicropores (cm3/g)

0.0128

Mean pore diameter (nm) Apparent density (g/cm3) Intraparticle porosity (%) Total porosity (%) Real density (g/cm3)

20.1 1.22 46.5 46.5 2.39

was recorded at equilibrium. For initial pH of 3.5 or 7.0, values in the range of 7.5–9.5 were finally measured.

3. Results and discussion Experimental results obtained in the adsorption runs are presented in Table 4. A quadratic model, given by Eq. (1), was used for preliminary regression fits (JMP 5.0.1 software). Linear, interaction and quadratic effects of the factors (in the coded levels x1, x2 and x3, see Table 3) were considered in this model. Coefficients of the respective effects were represented by β. q ¼ b0 þ b1 x1 þ b2 x2 þ b3 x3 þ b11 x21 þ b22 x22 þ b33 x23 þ b12 x1 x2 þ b13 x1 x3 þ b23 x2 x3 :

ð1Þ

These preliminary fittings were done to determine the significant effects on the responses for each dye. The effects determined as non-significant were then eliminated and a new fit was performed, resulting in a reduced model. Experimental data obtained for Astrazon Red adsorption demonstrate that the interaction term Cin–T (coded by x1 · x3) and the quadratic term T2 (x32) were non-significant. Thus, a reduced model, represented by Eq. (2), was used to fit experimental data of Astrazon Red adsorption by sepiolite: q ¼ b0 þ b1 x1 þ b2 x2 þ b3 x3 þ b11 x21 þ b22 x22 þ b12 x1 x2 þ b23 x2 x3 :

ð2Þ

The preliminary fit obtained for Sirius Blue adsorption demonstrates that the linear effect of temperature (x3) and the interaction effects Cin–T (x1 · x3) and pH–T (x2 · x3) were nonsignificant. Although non-significant, linear effect of tempera-

Table 3 Experimental range and coded values of the factors Dye

Astrazon Red

Sirius Blue

139

Run

1a 1b 2a 2b 3a 3b 4a 4b 5a 5b 6a 6b 7a 7b 8a 8b 9a 9b 10a 10b 11a 11b 12a 12b 13a 13b 14a 14b 15a 15b

x1 −1 −1 −1 −1 −1 −1 −1 −1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1

x2 −1 −1 0 0 0 0 1 1 −1 −1 −1 −1 0 0 0 0 0 0 1 1 1 1 −1 −1 0 0 0 0 1 1

x3

0 0 −1 −1 1 1 0 0 −1 −1 1 1 0 0 0 0 0 0 −1 −1 1 1 0 0 −1 −1 1 1 0 0

Astrazon Red

Sirius Blue

qexp (mg/g)

qpred (mg/g)

qexp (mg/g)

qpred (mg/g)

58.06 59.68 62.47 62.36 62.73 62.48 62.41 62.30 82.44 80.03 75.57 76.74 91.40 92.72 91.07 90.26 91.82 92.63 88.74 90.68 104.8 103.1 86.83 82.98 99.60 101.5 103.4 105.9 101.08 101.58

55.34 55.34 62.31 62.31 65.68 65.68 62.92 62.92 83.47 83.47 77.18 77.18 92.21 92.21 92.21 92.21 92.21 92.21 87.85 87.85 100.9 100.9 85.18 85.18 98.61 98.61 102.0 102.0 105.7 105.7

85.99 85.06 78.47 75.47 83.27 78.08 5.24 9.94 172.6 174.5 154.6 155.8 140.0 141.6 138.3 143.1 142.5 141.0 21.66 15.99 11.49 10.81 254.4 253.5 142.44 146.0 174.3 160.9 11.70 9.47

84.87 84.87 76.54 76.54 76.79 76.79 12.56 12.56 167.1 167.1 167.3 167.3 141.1 141.1 141.1 141.1 141.1 141.1 12.04 12.04 12.30 12.30 249.0 249.0 157.9 157.9 158.2 158.2 11.24 11.24

ture could not be eliminated from the model to support hierarchy (quadratic effect of temperature is significant). The interaction effects were then eliminated and a new regression was performed for Sirius Blue adsorption, based on Eq. (3): q ¼ b0 þ b1 x1 þ b2 x2 þ b3 x3 þ b11 x21 þ b22 x22 þ b33 x23 þ b12 x1 x2

ð3Þ

Quadratic reduced model fits were then done for adsorption of Astrazon Red and Sirius Blue using Eqs. (2) and (3), respectively. Table 4 shows the values of the responses predicted by these reduced models. Their adequacy was further analyzed by the correlation coefficient values, analysis of variance, lackof-fit tests and residuals analysis. 3.1. Correlation coefficients

Independent variables

Coded values (x) −1

0

1

Cin (mg/L) pHin T (°C) Cin (mg/L) pHin T (°C)

65 2 20 50 3.5 20

120 5 30 100 7 30

175 8 40 150 10.5 40

The regression coefficient (R2) quantitatively evaluates the correlation between the experimental data and the predicted responses. The obtained R2 values suggest good adjustments to the experimental results since these indicate that 97.7% (for Astrazon Red) and 99.1% (for Sirius Blue) of the variability in the response could be explained by the models. Adjusted R2 (Adj R2) is also a measure of goodness of a fit, but it's more

140

S.C.R. Santos, R.A.R. Boaventura / Applied Clay Science 42 (2008) 137–145

Table 5 ANOVA and lack-of-fit test of the selected quadratic model for Astrazon Red adsorption by sepiolite Sources of variation

Degrees of freedom

Analysis of variance Model 7 Error 22 Corrected total 29

Sum of squares

Mean square

7256.9641 169.2789 7426.2430

1036.71 7.69

F-ratio

Prob N F

134.7339 b0.0001

3.3. Lack-of-fit test

Lack-of-fit test Lack of fit 5 Pure error 17 Total error 22

144.29237 24.98655 169.27892

28.8585 1.4698 19.6343

b0.0001

R2 = 0.977; Adj R2 = 0.970.

suitable for comparing models with different numbers of independent variables. It corrects the R2-value for the sample size and the number of terms in the model by using the degrees of freedom on its computations, so if there are many terms in a model and not very large sample size, Adjusted R2 may be visibly smaller than R2 (Liu and Chiou, 2005). Here, Adjusted R2 values (0.970 for Astrazon Red and 0.988 for Sirius Blue) were very close to the corresponding R2 values. 3.2. Analysis of variance Analysis of variance (ANOVA) tests the significance and the adequacy of the regression model (Liu and Chiou, 2005) and is presented in Tables 5 and 6. ANOVA subdivides the total variation of the results in two components: variation associated with the model and variation associated with the experimental error, showing whether the variation from the model is significant or not when compared with the ones associated with residual error (Segurola et al., 1999). This comparison is performed by the F-value, which is the ratio between the mean square of the model and the residual error. If the model is a good predictor of the experimental results, F-value should be greater than the tabulated value (Mahat et al., 2004) of the F-distribution for a certain number of degrees of freedom in the model at a level of significance α. F-ratios obtained for

Table 6 ANOVA and lack-of-fit test of the selected quadratic model for Sirius Blue adsorption by sepiolite Sources of variation

Degrees of Sum of freedom squares

Analysis of variance Model 9 Error 20 Corrected total 29 Lack-of-fit test Lack of fit 5 Pure error 17 Total error 22 R2 = 0.991; Adj R2 = 0.988.

Astrazon Red and Sirius Blue adsorption, 134.7 and 354.9 respectively, are clearly greater than the tabulated F (2.5 at 95% significance) confirming the adequacy of the model fits. Prob N F is the probability that all the variation in the results are due to random error (Segurola et al., 1999), and thus the very low values obtained for the two dyes (b0.0001) indicate that results are not random and the terms in the models have a significant effect in the response (Noordin et al., 2004).

148,999.39 1319.62 150,319.01

1157.1594 162.4655 1319.6249

Mean square 21,285.6 60.0

F-ratio

Prob N F

354.8613 b0.0001

231.432 9.557 24.2165

b0.0001

Lack of fit is a special diagnostic test for adequacy of a model that compares the pure error, based on the replicate measurements, and other the lack of fit, based on the model performance (Noordin et al., 2004). F-value, calculated as the ratio between the lack-of-fit mean square and the pure error mean square, is the statistic parameter used to determine whether the lack of fit is significant or not, at a significance level α. In the two studied systems, Prob N F values were b 0.0001, revealing an undesirable significant lack of fit. 3.4. Residuals analysis In addition to these tests, the adequacy of the models was also evaluated by the residuals (difference between the experimental and the predicted values). Residuals are thought as elements of variation unexplained by the fitted model and then it is expected that they occur according to a normal distribution (NIST/SEMATECH). Normal probability plots are a suitable graphical method for judging the normality of the residuals. The observed residuals are plotted against the theoretical values, given by a normal distribution (see Fig. 1). The approximate straight lines obtained indicate that residuals are normally distributed. Residuals should also present structureless patterns (Noordin et al., 2004) when plotted against predicted values (Fig. 2), showing no increase as the size of the fitted value increases. Trends observed in Figs. 1 and 2, reveal reasonably well-behaved residuals. 3.5. Models and effect of the factors Tables 7 and 8 present the model parameter estimates for the adsorption of the two dyes. Student “t” test was used to verify the significance of the regression coefficients, testing whether the true parameter is zero or not. The larger the magnitude of t-value and smaller the Prob N |t|, the more significant is the corresponding coefficient. For both dyes, the low Prob N |t| values indicate significance of all the terms included (nonsignificant terms were already excluded in the preliminary fits, as they presented high values of Prob N |t|). For Astrazon Red adsorption, it could be concluded that the linear effect of initial concentration was the most significant effect. Initial pH and the quadratic effect of initial concentration were also very significant. Linear effect of temperature showed the lowest significance among the terms included. For Sirius Blue adsorption, initial concentration was also very significant, but initial pH was clearly, the most significant factor.

S.C.R. Santos, R.A.R. Boaventura / Applied Clay Science 42 (2008) 137–145

141

Table 7 Regression model results for Astrazon Red

Fig. 1. Normal probability plot of standardised residuals for (a) Astrazon Red and (b) Sirius Blue adsorption by sepiolite.

Interaction effect of these two factors showed also significance as well as quadratic effects of each one of the three factors. The final equations of the response dependence on the

Term

Coefficient

Standard error

t-ratio

Prob N |t|

Intercept x1 x2 x3 x1 · x2 x2 · x3 x1 · x3 x21 x22 x23

92.21 18.15 7.023 1.684 3.235 4.831 – − 10.06 − 4.860 –

0.9422 0.6935 0.6935 0.6935 0.9807 0.9807 – 1.018 1.018 –

97.9 26.2 10.1 2.43 3.30 4.93 – − 9.89 − 4.78 –

b0.0001 b0.0001 b0.0001 0.0238 0.0033 b0.0001 – b0.0001 b0.0001

three studied factors are given by Eq. (4) for Astrazon Red and Eq. (5) for Sirius Blue. Cin 210 pH 5 q ¼ 92:21 þ 18:15 þ 7:023 50 3 T 30 Cin 210 þ 1:684 þ 3:235 50 10 pH 5 pH 5 T 30 þ 4:831 3 3 10 2 Cin 210 pH 5 2 10:06 4:860 ð4Þ 50 3 Cin 100 pH 7 q ¼ 141:1 þ 40:70 77:51 3:5 50 T 30 Cin 100 þ 0:1288 41:35 10 50 2 pH 7 Cin 100 39:68 12:00 3:5 50 2 2 pH 7 T 30 11:73 ð5Þ 3:5 10 Astrazon Red adsorption model predicts a minimum adsorbed amount of 52.2 mg/L (for Cin = 65 mg/L, pHin = 2.0 and T = 40 °C) and a maximum amount of 112 mg/g (Cin = 175 mg/L, pHin = 8.0 and T = 40 °C). Inside the studied data range, sepiolite showed considerable capacities for this basic dye. It's known that monovalent organic cations may bind to a clay mineral surface by three possibilities: (1) an electrostatic interaction between the organic (or inorganic) cation and a Table 8 Regression model results for Sirius Blue

Fig. 2. Residuals versus predicted response for (a) Astrazon Red and (b) Sirius Blue adsorption by sepiolite.

Term

Coefficient

Standard error

t-ratio

Prob N |t|

Intercept x1 x2 x3 x1 · x2 x2 · x3 x1 · x3 x21 x22 x23

141.1 40.70 − 77.51 0.1288 − 41.35 – – − 12.00 − 39.68 − 11.73

1.117 0.6839 0.6839 0.6839 0.9671 0.9671 0.9671 1.007 1.007 1.007

44.6 21.0 − 40.0 0.070 − 15.1 – – − 4.21 − 13.9 − 4.12

b0.0001 b0.0001 b0.0001 0.9476 b0.0011 – – 0.0004 b0.0001 0.0005

142

S.C.R. Santos, R.A.R. Boaventura / Applied Clay Science 42 (2008) 137–145

monovalent negative site on the silicate blocks, resulting in a neutral complex; (2) a second organic cation may bind to a neutral sepiolite–organic complex by noncoulombic interactions, forming a single positively charged complex with two organic cations and one charged site; (3) in sepiolite, where neutral sites occur at the external surface, a monovalently charged complex may form by the binding of one organic cation and a neutral site (Bilgic, 2005). Measured cationic exchange capacity of sepiolite indicates that a maximum of 96 mg of Astrazon Red dye/g sepiolite could be adsorbed by ion exchange. This value was exceeded for some conditions of pH and temperature, indicating that adsorption also takes place on neutral sites and complexes. Similar result was found in the adsorption of the cationic dye Methylene Blue onto clay minerals (Gurses et al., 2006). It was also reported that clay minerals with adsorbed amounts of organic cations exceeding the CEC move to the negative electrode (Margulies et al., 1988), proving the formation of positively charged complexes. Sirius Blue adsorption model predicts a maximum capacity of 249 mg/g (for Cin = 150 mg/L, pHin = 3.5, T = 30 °C) which becomes almost insignificant at initial pH of 10.5 for all the initial concentrations and temperatures used. Figs. 3 and 4 show the response surface plots obtained for the adsorption of the two studied dyes. Fig. 3(a) illustrates the adsorption capacity dependence on initial pH and temperature for constant initial dye concentration. For Astrazon Red adsorption, at initial concentration of 120 mg/L, a significant increase in the amount adsorbed at equilibrium with the increase of initial solution pH from 2 to 8 can be observed. This expected observation is a result of the increase in the negative surface charge on the adsorbent as the pH increases, leading to a higher degree of cationic species adsorption (Wang et al., 2004). There was however an optimum initial pH that maximizes the adsorbed amount. For high temperature and dye concentration, it seems that the optimum initial pH should be above 8 (outside the studied data range). The lower the initial concentration and the temperature, the lower the optimum pH is. Some studies with other basic dyes and adsorbents (Ravikumar et al., 2005a,b; Wang et al., 2004) reported increase in adsorption capacity with pH, but others reported the existence of an optimum value (Gurses et al., 2006). In Fig. 3(a), it can also be observed that for low pH values, a decrease in temperature favors adsorption on sepiolite (exothermic process), suggesting a physical sorption based on weakening forces; for high pH, the adverse effect occurs, since an endothermic behaviour is observed, suggesting that chemical forces could be responsible for adsorption at high pH. In agreement, Fig. 3(b) shows that at initial pH 8, an increase in temperature between 20 and 40 °C causes a linear slight increase in the adsorbed amount for all initial dye concentrations. Literature reports endothermic nature for adsorption of the basic dyes Astrazon Blue FGRL (Karagozoglu et al., 2007) and Methylene Blue (Dogan et al., 2007) onto sepiolite in acidic conditions. Nevertheless, the temperature effect has been usually studied at the most favorable pH conditions and no information about exothermic or endothermic adsorption nature has been found for basic dyes in acidic conditions. The strong effect

Fig. 3. Response surface for Astrazon Red adsorption by sepiolite for (a) constant initial dye concentration of 120 mg/L, (b) initial solution pH 8 and (c) temperature 30 °C.

of initial concentration in Astrazon Red adsorbed amount is clearly visible in Fig. 3(a) and (b). For Sirius Blue adsorption, the strong effect of initial pH is illustrated for different temperatures and for an initial dye concentration of 100 mg/L in Fig. 4(a). The adsorption capacity of Sirius Blue decreases when pH increases to strong alkaline conditions as a result of repulsion between the negative charged surface and the anionic direct dye. As seen above, temperature linear effect is not significant but has a significant quadratic effect (Fig. 4(a) and (b)). The slight curvatures in the plots of q versus T at constant initial dye concentration indicate the

S.C.R. Santos, R.A.R. Boaventura / Applied Clay Science 42 (2008) 137–145

143

varied inside the experimental data range and equilibrium adsorbed amount calculated by Eqs. (4) and (5). Equilibrium dye concentrations in the liquid phase were then calculated by mass balance. Fig. 5 presents the adsorption equilibrium isotherms at different conditions (best and worst conditions and common conditions of neutral pH and 20 °C of temperature) obtained by the Response Surface Methodology. These data were additionally used for fitting the well-known Langmuir and Freundlich models for adsorption equilibrium. Langmuir isotherm (Langmuir, 1918) is represented by: qeq ¼

Qmax KL Ceq 1 þ KL Ceq

ð6Þ

where Qmax is the maximum adsorption capacity, corresponding to a monolayer coverage, and KL the Langmuir constant related to the energy of adsorption. Freundlich isotherm (Freundlich, 1906) is represented by: 1=n qeq ¼ KF Ceq

ð7Þ

where KF and 1/n are model parameters related to adsorption capacity and adsorption intensity, respectively. Data obtained by RSM were then fitted to Eqs. (6) and (7) by nonlinear regression, using the software Fig.P from Biosoft. Predicted

Fig. 4. Response surface for Sirius Blue adsorption by sepiolite (a) constant initial dye concentration of 100 mg/L, (b) initial solution pH 7, and (c) temperature 30 °C.

existence of an optimum inside the temperature range of 20– 40 °C. Literature reports an endothermic adsorption of the anionic dye Reactive Blue 221 onto sepiolite, in the range of 20–50 °C (Alkan et al., 2007), but an exothermic nature has been reported for Acid Blue 193 adsorption, in the temperature range of 20–40 °C (Ozcan et al., 2006). 3.6. Equilibrium adsorption isotherms The predicted models obtained for the adsorption of these two dyes were furthermore used to plot isotherms at different pH and temperature conditions. Initial dye concentrations were

Fig. 5. Adsorption equilibrium isotherms for (a) Astrazon Red and (b) Sirius Blue: experimental, RSM results and Langmuir (──) and Freundlich () modelling.

144

S.C.R. Santos, R.A.R. Boaventura / Applied Clay Science 42 (2008) 137–145

Table 9 Langmuir and Freundlich isotherms parameters for adsorption of Astrazon Red and Sirius Blue Dye

T pHin Langmuir (°C) Qmax KL (mg/g) (L/mg)

Astrazon 20 Red 40 40 Sirius 20 Blue 30 30

7.0 2.0 8.0 7.0 3.5 10.5

106 93.6 108 202 454 55.2

0.197 0.101 1.39 0.0506 0.0523 0.00858

Freundlich 2

R

KF n (mg/g/(mg/L)1/n)

1.00 5.12 1.00 4.69 0.98 10.8 1.00 2.63 0.94 1.71 0.96 1.56

44.9 33.2 76.9 32.3 40.1 1.36

R2 0.99 0.98 1.00 0.99 0.93 0.96

curves and respective parameters are presented in Fig. 5 and Table 9 respectively. 4. Conclusions Adsorption modelling of Astrazon Red and Sirius Blue dyes by sepiolite as a function of initial dye concentration, initial pH and temperature was performed by using a Box–Behnken design. Sepiolite showed high adsorption capacities for Astrazon Red basic dye over all the range of the experimental conditions tested. Predicted adsorption capacities varied between 52.2 mg/g (for Cin = 65 mg/L, pHin = 2.0 and T = 40 °C) and 112 mg/g (for Cin = 175 mg/L, pHin = 8.0 and T = 40 °C). Sirius Blue adsorption showed to be very dependent on the initial solution pH. Maximum predicted adsorption capacity was quite great, 249 mg/g (for Cin = 150 mg/L, pHin = 3.5, T = 30 °C), but is almost insignificant at initial pH of 10.5. Acknowledgements The authors are grateful to the Portuguese Foundation for Science and Technology (FCT) for the financial support and S. Santos' doctorate scholarship (SFRH/BD/18477/2004). References Acemioglu, B., 2004. Adsorption of Congo red from aqueous solution onto calcium-rich fly ash. Journal of Colloid and Interface Science 274 (2), 371–379. Ahmadi, M., Vahabzadeh, F., Bonakdarpour, B., Mofarrah, E., Mehranian, M., 2005. Application of the central composite design and response surface methodology to the advanced treatment of olive oil processing wastewater using Fenton's peroxidation. Journal of Hazardous Materials 123 (1–3), 187–195. Aksu, Z., Tezer, S., 2000. Equilibrium and kinetic modelling of biosorption of Remazol Black B by Rhizopus arrhizus in a batch system: effect of temperature. Process Biochemistry 36 (5), 431–439. Alkan, M., Demirbas, O., Dogan, M., 2007. Adsorption kinetics and thermodynamics of an anionic dye onto sepiolite. Microporous and Mesoporous Materials 101 (3), 388–396. Alrichs, J.L., Serna, J.C., Serratosa, J.M., 1975. Structural hydroxyls in sepiolites. Clays and Clay Minerals 23, 119–124. Annadurai, G., Juang, R.S., Lee, D.J., 2002. Factorial design analysis for adsorption of dye on activated carbon beads incorporated with calcium alginate. Advances in Environmental Research 6 (2), 191–198.

Annadurai, G., Sheeja, R.Y., 1998. Use of Box–Behnken design of experiments for the adsorption of verofix red using biopolymer. Bioprocess Engineering 18 (6), 463–466. Bhatnagar, A., Jain, A.K., 2005. A comparative adsorption study with different industrial wastes as adsorbents for the removal of cationic dyes from water. Journal of Colloid and Interface Science 281 (1), 49–55. Bilgic, C., 2005. Investigation of the factors affecting organic cation adsorption on some silicate minerals. Journal of Colloid and Interface Science 281 (1), 33–38. Box, G.E.P., Behnken, D.W., 1960. Some new three level designs for the study of quantitative variables. Technometrics 2, 455–475. Catalkaya, E.C., Kargi, F., 2007. Effects of operating parameters on advanced oxidation of diuron by the Fenton's reagent: a statistical design approach. Chemosphere 69 (3), 485–492. Dogan, M., Ozdemir, Y., Alkan, M., 2007. Adsorption kinetics and mechanism of cationic methyl violet and methylene blue dyes onto sepiolite. Dyes and Pigments 75 (3), 701–713. Figueiredo, S.A., Loureiro, J.M., Boaventura, R.A., 2005. Natural waste materials containing chitin as adsorbents for textile dyestuffs: batch and continuous studies. Water Research 39 (17), 4142–4152. Freundlich, H.M.F., 1906. Over the adsorption in solution. Journal of Physical Chemistry 57, 385–471. Garg, V.K., Gupta, R., Yadav, A.B., Kumar, R., 2003. Dye removal from aqueous solution by adsorption on treated sawdust. Bioresource Technology 89 (2), 121–124. Gurses, A., Dogar, C., Yalcin, M., Acikyildiz, M., Bayrak, R., Karaca, S., 2006. The adsorption kinetics of the cationic dye, methylene blue, onto clay. Journal of Hazardous Materials 131 (1–3), 217–228. Kadirvelu, K., Kavipriya, M., Karthika, C., Radhika, M., Vennilamani, N., Pattabhi, S., 2003. Utilization of various agricultural wastes for activated carbon preparation and application for the removal of dyes and metal ions from aqueous solutions. Bioresource Technology 87 (1), 129–132. Karagozoglu, B., Tasdemir, M., Demirbas, E., Kobya, M., 2007. The adsorption of basic dye (Astrazon Blue FGRL) from aqueous solutions onto sepiolite, fly ash and apricot shell activated carbon: kinetic and equilibrium studies. Journal of Hazardous Materials 147 (1–2), 297–306. Langmuir, I., 1918. The adsorption of gases on plane surfaces of glass, mica and platinum. Journal of the American Chemical Society 40, 1361–1403. Liu, H.L., Chiou, Y.R., 2005. Optimal decolorization efficiency of Reactive Red 239 by UV/TiO2 photocatalytic process coupled with response surface methodology. Chemical Engineering Journal 112 (1–3), 173–179. Mahat, M.K., Illias, R.M., Rahman, R.A., Rashid, N.A.A., Mahmood, N.A.N., Hassan, O., Aziz, S.A., Kamaruddin, K., 2004. Production of cyclodextrin glucanotransferase (CGTase) from alkalophilic Bacillus sp TS1-1: media optimization using experimental design. Enzyme and Microbial Technology 35 (5), 467–473. Margulies, L., Rozen, H., Nir, S., 1988. Model for competitive adsorption of organic cations on clays. Clays and Clay Minerals 36 (3), 270–276. Martin, M.J., Artola, A., Balaguer, M.D., Rigola, M., 2003. Activated carbons developed from surplus sewage sludge for the removal of dyes from dilute aqueous solutions. Chemical Engineering Journal 94 (3), 231–239. Mohapatra, H., Gupta, R., 2005. Concurrent sorption of Zn(II), Cu(II) and Co(II) by Oscillatoria angustissima as a function of pH in binary and ternary metal solutions. Bioresource Technology 96 (12), 1387–1398. Montgomery, D.C., 1991. Design and Analysis of Experiments. New York. Netpradit, S., Thiravetyan, P., Towprayoon, S., 2004. Evaluation of metal hydroxide sludge from reactive dye adsorption in a fixed-bed column system. Water Research 38 (1), 71–78. NIST/SEMATECH e-Handbook of Statistical Methods. http://www.itl.nist.gov/ div898/handbook/pri/pri.htm, access date: 2nd November, 2007. Noordin, M.Y., Venkatesh, V.C., Sharif, S., Elting, S., Abdullah, A., 2004. Application of response surface methodology in describing the performance of coated carbide tools when turning AISI 1045 steel. Journal of Materials Processing Technology 145 (1), 46–58. Ozcan, A.S., Ozcan, A., 2004. Adsorption of acid dyes from aqueous solutions onto acid-activated bentonite. Journal of Colloid and Interface Science 276 (1), 39–46.

S.C.R. Santos, R.A.R. Boaventura / Applied Clay Science 42 (2008) 137–145 Ozcan, A., Ozcan, A.S., 2005. Adsorption of Acid Red 57 from aqueous solutions onto surfactant-modified sepiolite. Journal of Hazardous Materials 125 (1–3), 252–259. Ozcan, A.S., Erdem, B., Ozcan, A., 2005. Adsorption of Acid Blue 193 from aqueous solutions onto BTMA-bentonite. Colloids and Surfaces. A, Physicochemical and Engineering Aspects 266 (1–3), 73–81. Ozcan, A., Oncu, E.M., Ozcan, A.S., 2006. Kinetics, isotherm and thermodynamic studies of adsorption of Acid Blue 193 from aqueous solutions onto natural sepiolite. Colloids and Surfaces. A, Physicochemical and Engineering Aspects 277 (1–3), 90–97. Ozdemir, O., Armagan, B., Turan, M., Celik, M.S., 2004. Comparison of the adsorption characteristics of azo-reactive dyes on mesoporous minerals. Dyes and Pigments 62 (1), 49–60. Ravikumar, K., Deebika, B., Balu, K., 2005a. Decolourization of aqueous dye solutions by a novel adsorbent: application of statistical designs and surface plots for the optimization and regression analysis. Journal of Hazardous Materials 122 (1–2), 75–83. Ravikumar, K., Pakshirajan, K., Swaminathan, T., Balu, K., 2005b. Optimization of batch process parameters using response surface methodology for dye removal by a novel adsorbent. Chemical Engineering Journal 105 (3), 131–138.

145

Robinson, T., Chandran, B., Nigam, P., 2002. Removal of dyes from an artificial textile dye effluent by two agricultural waste residues, corncob and barley husk. Environment International 28 (1–2), 29–33. Segurola, J., Allen, N.S., Edge, M., Mc Mahon, A., 1999. Design of eutectic photoinitiator blends for UV/visible curable acrylated printing inks and coatings. Progress in Organic Coatings 37 (1–2), 23–37. Silva, J.P., Sousa, S., Goncalves, I., Porter, J.J., Ferreira-Dias, S., 2004. Modelling adsorption of acid orange 7 dye in aqueous solutions to spent brewery grains. Separation and Purification Technology 40 (2), 163–170. Sun, Q.Y., Yang, L.Z., 2003. The adsorption of basic dyes from aqueous solution on modified peat-resin particle. Water Research 37 (7), 1535–1544. Walker, G.M., Weatherley, L.R., 1999. Kinetics of acid dye adsorption on GAC. Water Research 33 (8), 1895–1899. Wang, C.C., Juang, L.C., Hsu, T.C., Lee, C.K., Lee, J.F., Huang, F.C., 2004. Adsorption of basic dyes onto montmorillonite. Journal of Colloid and Interface Science 273 (1), 80–86. Wibulswas, R., 2004. Batch and fixed bed sorption of methylene blue on precursor and QACs modified montmorillonite. Separation and Purification Technology 39 (1–2), 3–12.