Adsorption of safranin-O dye on CO₂ neutralized activated red mud waste: process modelling, analysis and optimization using statistical design

Abstract

Response surface methodology (RSM) was applied to examine the removal of hazardous safranin-O dye from aqueous solution by CO₂ neutralized activated red mud. A 2⁴ full factorial central composite design (CCD) was used to evaluate the effects of adsorption parameters such as, adsorbent dose (X₁), temperature (X₂), solution pH (X₃) and initial safranin-O concentration (X₄) on percentage removal of safranin-O dye from aqueous solution. Analysis of variance (ANOVA) exhibit a high R² value of 0.9655, indicating second-order regression model excellently evaluate the experimental data. The interaction effects of the main factor and optimum condition of the process were determined from contour and response surface plots, respectively. In addition, the isotherm study revealed that, the adsorption data were best fitted to Langmuir model with high correlation value of 0.994 and adsorption capacity of 9.7680 mg g⁻¹. The optimum operating condition for adsorption of safranin-O dye were obtained as adsorbent dose (0.62 g), temperature (29.06 °C), pH (8.3) and initial safranin-O concentration (37.3 mg L⁻¹). At the optimum condition, the adsorption of safranin-O from aqueous solution was found to be 94.5 ± 0.1%.

---

1. Introduction

Dyes are usually used to colour materials such as leather, plastics, textiles, food, paper, and cosmetics in many industries.¹ Among various industries, the textile industry uses large variety of dyes for coloration of fibre. As a result, a substantial amount of coloured wastewater was generated from these industries which create serious environmental problem not only because of its toxicity, but also due to its visibility problem.^2–4 Therefore, the dyes should be removed from industrial effluent before discharge to the environment is an important aspect for both artistic sense and health point of view.

Various conventional methods were used for the removal of dyes from the industrial waste water.^5–12 Among them adsorption process is preferred widely to treat wastewater containing different class of dyes. Recently, many low-cost and effective adsorbent have been used by the researchers for the adsorption of dye from water.^13–19 However, some disadvantage were associated with the adsorbent like, regeneration of the adsorbent is expensive, not straightforward and result in loss of the adsorbent. Due to the above problems research interest into the production of alternative adsorbent to minimize the disadvantages.

Red mud is one of the main waste material in alumina industries worldwide. The quantity of red mud generated, depends upon the quality of bauxite processed. A huge amount of red mud released by these industries is an economical and environmental problem which has led researchers to develop new uses of red mud.²⁰ Red mud is highly alkaline with a pH > 12 and requires neutralization. Neutralisation of red mud to pH around 8.0 is optimal because the chemically adsorbed Na is released, alkaline buffer minerals are neutralized and toxic metals are insoluble at this pH.²¹ The global climate change due to increase in anthropogenic carbon dioxide (CO₂) concentrations in the atmospheric air constructing a challenging issue for future generations.²² The energy sector is a major contributor of CO₂ emissions, with estimates placing their current levels above 30 Gt.²³ The issues of climate change can be minimize by reducing the carbon emission from the industries and use it by capturing and storing. Thus, in this study, carbon dioxide gas was used to neutralize the alkaline red mud. Also, utilization of industrial wastes as an assets to resolve the problem of another wastes provide commercial profit.

Among different class (anionic, cationic and non-ionic) of dyes, cationic dye (safranin-O) is chosen as the target because it is more toxic and has harmful effects on living organisms during short period of exposure. Table 1 listed some of the reported biosorbent used for removal of safranin-O from aqueous solutions. However, to the best of our knowledge, the applications of red mud on the adsorption of safranin-O has not been reported before. Therefore, it will be significant to study the interaction between the efficient, cheap and easily available red mud and safranin-O dye.

Table 1 Adsorption capacities of safranin-O dyes onto different adsorbents

Adsorbent(s)	pH	Removal capacity (%)
Calcined mussel shells^24	>9.2	87.56
Pineapple peels^25	6–8	43.3
Calcined bones^15	6.2	96.78
NaOH treated rice husk^26	8	98.02
Alkali-treated rice husk^27	8	93.28
Corncob activated carbon^28	5–9	99.8
Present study	8.3	95.5

---

The aim of this study was to evaluate the removal efficiency of hazardous safranin-O from aqueous solution by using CO₂ neutralized activated red mud. In this study, three-level, four-factor central composite design (CCD) was employed to determine the relationship between experimental variables (adsorbent dose, temperature, pH and initial safranin-O concentration) and optimum adsorption conditions for activated red mud. Meanwhile, characteristics of the adsorption isotherm was studied through the adsorption experiments.

2. Experimental procedure

2.1. Materials

All the chemicals used in this investigation were Analytical grade, obtained from Merck, Mumbai (India) and Rankem, New Delhi (India). Safranin-O dye was purchased from Sigma-Aldrich, Steinheim (Germany). Stock solution (1000 ppm) of safranin-O was prepared by dissolving 1 g in distilled water (1 L). Required standard solutions were prepared from the stock solution by dilution and their concentration were confirmed using calibration curve. pH of the safranin-O solution was adjusted by adding drops of 0.1 M NaOH or 0.1 M HCl solution.

2.2. Preparation of adsorbent

Red mud obtained from the Vedanta Aluminium Limited, Lanjigarh, Odisha (India) was used in this study. The chemical compositions of the red mud on the dry weight are Fe₂O₃ (54%), Al₂O₃ (13%), SiO₂ (7%), Na₂O (8%), and TiO₂ (3.5%). Starting the red mud was separated from alkaline solution by decanting. Approximately 10 g red mud was mixed with 100 mL distilled water in a 500 mL gas close-fitting plastic bottle with a bay for CO₂ gas and an opening to escape the pressure. The solution mixture was fitted with a mechanical stirrer adjusted to a constant stirring speed (180 rpm) at room temperature. The CO₂ gas passed through a rotameter and mixed into the red mud solution with a gas flow rate of 5 mL min⁻¹. The CO₂ gas was passed until the pH of the red mud suspension was decreased from 12.8 to 7.37. The neutralized red mud was separated by centrifugation for 20 min at 3000 rpm and dried in an air oven at 110 °C for 2 h. The material was calcined at 500 °C for 2 h which referred as CO₂ neutralised activated red mud (ARM) and stored in an airtight glass bottle until used.

2.3. Analysis

The surface area and particle size of the adsorbent were measured by BET method (Quanta chrome AUTOSORB-1, USA) and particle size analyser (Nano-ZS 90, MALVERN, UK) respectively. The pH of all solutions were checked by using PHAN Lab India pH analyser. The concentration of safranin-O in water was determined using UV-visible spectrophotometer (Perkin-Elmer) at λ_max = 520 nm. A Perkin-Elmer RX-I FT-IR spectrometer was used to obtain a 4 cm⁻¹ resolution spectra in the range 400–4000 cm⁻¹ of the adsorbent before and after adsorption. The SEM images of ARM before and after adsorption of safranin-O were recorded by JEOL, JSM-6480LV analyser. The powder X-ray diffraction (XRD) of red mud was determined by using Philips X'Pert X-ray diffractometer.

2.4. Adsorption studies

Batch adsorption experiments were performed in 30 sets of 100 mL conical flasks containing 50 mL of safranin-O dye solution. Appropriate amount of CO₂ neutralized activated red mud was added to each conical flask and stirred in a rotary shaker at 140 rpm with temperature 25 °C. Batch experiments were conducted in order to investigate the effect of various independent process variables include adsorbent dosage (X₁), temperature (X₂), solution pH (X₃) and Initial dye concentration (X₄) based on the central composite design (CCD). After each run, the supernatant was separated by centrifugation. The concentration of safranin-O in the supernatant was measured by UV-visible spectrophotometer at λ_max = 520 nm. The removal efficiency (% removal) and uptake capacity (q_e (mg g⁻¹)) of safranin-O dye was obtained using following equation:where C_i and C_f are the initial and final equilibrium concentrations of safranin-O dye (mg L⁻¹), respectively. w is mass of the adsorbent in gram and v is volume of the solution in litter.

2.5. Response surface methodology (RSM) based central composite design (CCD)

Response surface methodology (RSM) is a mathematical and statistical techniques²⁹ used for evaluating the optimum operating conditions and regression model equation. In this study, the Central Composite Design (CCD), a standard form of RSM was applied to optimize the adsorption process for maximum adsorption of safranin-O dye. Design Expert 7.0 software was used for statistical data analysis. The independent variables (factors) used in this study were: adsorbent dosage (X₁), temperature (X₂), solution pH (X₃) and Initial dye concentration (X₄). Three levels were used for the each variables: the lower (−1), the higher (+1) and the central point (0). The actual and coded values of each variable are shown in Table 2. A 2⁴ full-factorial experimental design, with eight axial points and six replicates at the centre point was employed. Thus a total of 30 experiments were performed in this study.

Table 2 Factors and levels used in the central composite design study

Factors'	Factor code	Level of factors
		−1	0	+1
Adsorbent dose (mg g^−1)	X1	0.1	0.55	1
Temperature (°C)	X2	10	35	60
pH	X3	2	7	12
Initial concentration (mg L^−1)	X4	10	30	50

---

The following second-order polynomial equation was used to analyse the experimental results;

where Y is the predicted response (% removal), β_o is the offset term, β_i is the linear effect, β_ij is the first-order interaction, β_ii is the quadratic effect and X_i, X_j are the independent variables of the model.

Analysis of variance (ANOVA) was applied to the response and the corresponding parameters to analyse the modelled and finding the optimum level, also evaluate the statistical parameters by means of response surface methods.

3. Result and discussion

3.1. Adsorbent characterization

Particle size and BET-N₂ surface area of the activated red mud was in the range of 0.1–150 μm and 67.10 m² g⁻¹, respectively. SEM images of the CO₂ neutralised activated red mud (Fig. 1) showed that the surface view of the activated red mud was changed after the adsorption of dye and the pores were completely filled with safranin-O and appeared puffy due to heterogeneity.

Fig. 1 SEM images of CO₂ neutralized activated red mud (a) before adsorption and (b) after adsorption.

The XRD pattern of activated red mud before and after adsorption is shown in Fig. 2. The main phases identified in the CO₂ neutralised activated red mud were hematite (α-Fe₂O₃), goethite (α-FeO(OH)), gibbsite (γ-Al(OH)₃), anatase (TiO₂) quartz (SiO₂) and calcite (CaCO₃). Activated red mud after adsorption shows similar XRD patterns with activated red mud before adsorption. However, the intensity of diffraction peaks was increases conspicuously whereas calcite (CaCO₃) mineral peak intensity was vanish, this may be due to the phase transformation.

Fig. 2 XRD pattern of CO₂ neutralized activated red mud before and after adsorption of safranin-O dye.

Fig. 3 shows the FTIR spectra of the safranin-O dye, CO₂ neutralised activated red mud before and after safranin-O adsorption. CO₂ neutralised activated red mud showed a broad peak at 3087 cm⁻¹ and a weak peak at 1634 cm⁻¹, due to the stretching vibration of O–H and CO₃²⁻ groups respectively. After adsorption, some new absorption bands were occurred on dye loaded activated red mud. The bands at 1603 and 1634 cm⁻¹ were assigned to aromatic ring. The band at 1424 cm⁻¹ was due to –CH₃ bending vibration. The peaks at 1335 and 1338 cm⁻¹ represented the aromatic-N. However, these newly presented peaks appeared after adsorption of dye, indicated the dye adsorbed onto CO₂ neutralised activated red mud. Additionally, the band at 3292 cm⁻¹ is attributed to N–H group, which confirms the interaction between the dye ions and –OH group of the adsorbent to make hydrogen bond.³⁰ According to the FTIR spectra the adsorption of safranin-O dye on activated red mud (ARM), the mechanism proposed can be illustrated in Fig. 4.

Fig. 3 FTIR spectrum of safranin-O dye and CO₂ neutralised activated red mud before and after adsorption of dye.

Fig. 4 Schematic representation of adsorption mechanism of safranin-O dye onto activated red mud.

3.2. Model determination

The model statistics for each model was output by Design Expert 7.0 software. A quadratic model was suggested, even though the R² and adjacent R² values of quadratic model was lower than the cubic model. This is because the cubic model is aliased, which means that the effects of each variable become indistinguishable due to different signals. For linear model and 2FI model, the R² and adjacent R² values are found to be 0.1962, 0.0676 and 0.3303, −0.0222, respectively, which shows both the models are not adequate for the experimental data. Therefore the quadratic model (R² = 0.9655 and adjacent R² = 0.9333) was selected to fit the experimental data. The experimental design matrix and their results are represented in Table 3.

Table 3 Experimental design matrix and response

Run	Coded values				Actual values				% Removal		Residual
	X1	X2	X3	X4	X1	X2	X3	X4	Observed	Predicted
1	−1	−1	−1	−1	0.1	10	2	10	59.68	63.13	−3.45
2	1	−1	−1	−1	1	10	2	10	76.26	75.47	0.79
3	−1	1	−1	−1	0.1	60	2	10	70.05	68.82	1.23
4	1	1	−1	−1	1	60	2	10	75.69	77.21	−1.52
5	−1	−1	1	−1	0.1	10	12	10	67.51	67.47	0.039
6	1	−1	1	−1	1	10	12	10	83.59	82.10	1.49
7	−1	1	1	−1	0.1	60	12	10	70.38	71.39	−1.01
8	1	1	1	−1	1	60	12	10	79.91	82.07	−2.16
9	−1	−1	−1	1	0.1	10	2	50	81.96	80.40	1.56
10	1	−1	−1	1	1	10	2	50	78.52	79.18	−0.66
11	−1	1	−1	1	0.1	60	2	50	73.01	76.17	−3.16
12	1	1	−1	1	1	60	2	50	70.35	71.00	−0.65
13	−1	−1	1	1	0.1	10	12	50	84.19	84.35	−0.16
14	1	−1	1	1	1	10	12	50	83.57	85.41	−1.84
15	−1	1	1	1	0.1	60	12	50	76.95	78.35	−1.40
16	1	1	1	1	1	60	12	50	77.24	75.46	1.78
17	−1	0	0	0	0.1	35	7	30	72.57	70.54	2.03
18	1	0	0	0	1	35	7	30	80.24	79.99	0.25
19	0	−1	0	0	0.55	10	7	30	76.52	76.55	−0.025
20	0	1	0	0	0.55	60	7	30	74.59	72.28	2.31
21	0	0	−1	0	0.55	35	2	30	73.44	71.66	1.78
22	0	0	1	0	0.55	35	12	30	80.96	80.47	0.49
23	0	0	0	−1	0.55	35	7	10	74.58	73.42	1.16
24	0	0	0	1	0.55	35	7	50	85.21	84.09	1.12
25	0	0	0	0	0.55	35	7	30	96.38	93.39	2.99
26	0	0	0	0	0.55	35	7	30	92.72	93.39	−0.67
27	0	0	0	0	0.55	35	7	30	92.18	93.39	−1.21
28	0	0	0	0	0.55	35	7	30	91.71	93.39	−1.68
29	0	0	0	0	0.55	35	7	30	92.35	93.39	−1.04
30	0	0	0	0	0.55	35	7	30	94.98	93.39	1.59

---

3.3. Development of regression model equation

CCD was used to build a relationship between four independent variables and removal of safranin-O from aqueous solution by second-order polynomial equation. Based on the model analysis, a quadratic model was chosen in terms of coded factor and actual factor for adsorption of safranin-O (% removal) and the equation obtained is shown in eqn (4):

Y = 93.3867 + 2.3642X₁ − 1.0650X₂ + 2.2025X₃ + 2.6658X₄ − 0.9875X₁X₂ + 0.5725X₁X₃ − 3.3913X₁X₄ − 0.4413X₂X₃ − 2.4800X₂X₄ − 0.1000X₃X₄ − 4.5304X₁X₁ − 4.7429X₂X₂ − 4.3317X₃X₃ − 3.6579X₄X₄ (4)

The positive sign and the negative sign of the term indicates the synergetic and antagonistic effect respectively.

Table 4 represents analysis of variance (ANOVA) results for safranin-O adsorption by activated red mud. Analysis of variance (ANOVA) was applied for the model to examine the fitness of the model, accuracy of the model, effects of single variables and interaction effects on the response.³¹ The model F value was found to be 29.97 with lower probability (<0.0001) indicates the model was significant. There was only a 0.01% possibility that the model F value could happen due to noise. It was observed that among the four variables studied, initial concentration (X₄) had the largest effect on the removal of safranin-O due to the maximum F value followed in order by adsorbent dose (X₁), pH (X₃) and time (X₂). The lack of fit for F value of 1.78 implies this terms is not significant relative to the pure error. There is only 27.20% chance that the lack of fit F value could occur due to noise.³² The predicted determination coefficient (R²) value is 0.9655.

Table 4 Analysis of variance (ANOVA) for the selected quadratic model^a

Source	Sum of squares	df	Mean square	F value	p-value prob > F
a R^2 = 0.9655, adj R^2 = 0.8339, # = significant, * = not significant.
odel	2206.50	14	157.61	29.97	<0.0001	#
X1	134.14	1	134.14	25.51	0.0001	#
X2	27.22	1	27.22	5.18	0.0380	#
X3	116.42	1	116.42	22.14	0.0003	#
X4	170.56	1	170.56	32.43	<0.0001	#
X1X2	15.60	1	15.60	2.97	0.1055	*
X1X3	5.24	1	5.24	1.00	0.3338	*
X1X4	184.01	1	184.01	34.99	<0.0001	#
X2X3	3.12	1	3.12	0.59	0.4535	*
X2X4	98.41	1	98.41	18.71	0.0006	#
X3X4	0.16	1	0.16	0.030	0.8639	*
X1X1	562.96	1	562.96	107.05	<0.0001	#
X2X2	617.01	1	617.01	117.33	<0.0001	#
X3X3	514.65	1	514.65	97.86	<0.0001	#
X4X4	367.00	1	367.00	69.79	<0.0001	#
Residual	78.88	15	5.26
Lack of fit	61.60	10	6.16	1.78	0.2720	*
Pure error	17.29	5	3.46
Cor total	2285.38	29

---

In Table 5, the estimated regression coefficients, t-values and P-values are given. Significance of each model term was checked using P-value. The P-values smaller than 0.05 indicates that the model is statistically significant, whereas the values greater than 0.1000 indicates the model terms are not significant. In this case all the linear model terms (X₁, X₂, X₃, and X₄) and quadratic model terms (X₁X₁, X₂X₂, X₃X₃ and X₄X₄) are significant, whereas only X₁X₄ and X₂X₄ are significant for the interaction model terms. Other variables such as X₁X₂, X₁X₃, X₂X₃ and X₃X₄ are not significant effect for the removal of safranin-O due to the p value more than 0.05.

Table 5 Estimated regression coefficients, t-values and P-values

Terms	Removal of safranin-O
	Coefficient	t-value	P-value
Constant	93.3867	99.750	0.000
X1	2.3642	5.051	0.000
X2	−1.0650	−2.275	0.038
X3	2.2025	4.705	0.000
X4	2.6658	5.695	0.000
X1X2	−0.9875	−1.722	0.106
X1X3	0.5725	0.999	0.334
X1X4	−3.3913	−5.915	0.000
X2X3	−0.4413	−0.770	0.453
X2X4	−2.4800	−4.326	0.001
X3X4	−0.1000	−0.174	0.864
X1X1	−4.5304	−10.346	0.000
X2X2	−4.7429	−10.832	0.000
X3X3	−4.3317	−9.893	0.000
X4X4	−3.6579	−8.354	0.000

---

Fig. 5a shows the comparison of predicted and experimental % removal of safranin-O dye, where the points bunch nearby the diagonal line, showing good fitness of the model, because the value of predicted R² of 0.8339 is in logical agreement with the adjusted R² of 0.9333. Fig. 5b shows the relationship between the normal probability (%) and the internally studentized residuals. The straight line means that neither response transformation was required nor apparent problem with normality.³³ Adequate precision measures the signal-to-noise ratio of 18.660 shows an adequate signal because when the ratio is greater than 4, the model is a desirable one.³⁴ Thus, this model can be used to navigate the design space.

Fig. 5 (a) Comparison of predicted and experimental % removal safranin-O dye, (b) normal plot of residuals showing the relationship between normal probability (%) and internally studentized residuals.

3.4. Model modification

After the evaluation of parameters significance, the model can be improved by excluding the terms that are not significant. The final model for describing the relationship between the adsorbent dose (X₁), temperature (X₂), pH (X₃) and initial concentration (X₄) is shown in eqn (5):

Y = 93.3867 + 2.3642X₁ − 1.0650X₂ + 2.2025X₃ + 2.6658X₄ − 3.3913X₁X₄ − 2.4800X₂X₄ − 4.5304X₁X₁ − 4.7429X₂X₂ − 4.3317X₃X₃ − 3.6579X₄X₄ (5)

After elimination of the non-significant term, the R² (0.9549) and adjusted R² (0.9312) values decreases slightly whereas the predicted R² and adequate precision increases from 0.8339 to 0.8808 and 18.660 to 20.783, respectively.

3.5. Effects of variables on the removal of safranin-O dye

To understand the interaction effects between the process variables and treatment outputs, the three dimensional (3D) response surface plots and there corresponding two dimensional (2D) counter plots of the model were constructed by using the statistical software (Fig. 6 and 7). The response surface plots better visualizes the interaction effects of each factor to influence the percentage removal of safranin-O dye from the aqueous solution. The shape of the contour plot shows the natures and extents of the interactions effects between the experimental factors on the response.³⁵ The circular and elliptical or saddle nature of the contour plots shows the negligible and significance interaction between the equivalent variables, respectively.³⁴ In each plot, the effects of two variables were varied within the experimental ranges and the other variable fixed to the zero level.

Fig. 6 The 3D plots and corresponding counter plots showing effect of (a) pH and temperature, (b) temperature and initial dye concentration, (c) temperature and adsorbent dose on percentage removal of safranin-O dye.

Fig. 7 The 3D plots and corresponding counter plots showing effect of (a) pH and initial dye concentration, (b) pH and adsorbent dose, and (c) initial dye concentration and adsorbent dose on percentage removal of safranin-O dye.

3.5.1. Effect of temperature and pH on adsorption process. The role of pH and temperature are considered to be the most influencing variables in adsorption process. The response surface and counter plot were developed to illustrate the combining effect of temperature and pH on the removal percentage of safranin-O at constant adsorbent dose of 0.55 mg g⁻¹ and initial concentration of 30 mg L⁻¹ (Fig. 6a). The figure shows that, the adsorption percentage of the safranin-O dye increases with increase in temperature from 10 °C to 30 °C. However, beyond this temperature percentage removal begins to decrease, which indicates that the adsorption of safranin-O onto the surface of adsorbent particles is favoured at lower temperatures and is controlled by an exothermic process. This may be due to the weakening of attractive forces between the dye molecules and the surface of particles.³⁶ In addition, the percentage removal of safranin-O increases with increase in pH of the solution from 2 to 10 and after pH 10 the percentage removal becomes constant. This is because, at higher pH, OH⁻ concentration on the adsorbent surface increases. Thus the electrostatic force of attraction between positively dye cation and negatively charged adsorbent sites increases.³⁷ By using a response optimizer, the predicted % removal, i.e., 93.73% was obtained at pH 8.8 and temperature 32.3 °C.

3.5.2. Effect of temperature and dye concentration on adsorption process. The combined effect of initial dye concentration and temperature on the dye adsorption onto activated red mud at constant pH 7 and adsorbent dose 0.55 mg g⁻¹ is shown in Fig. 6b. It was evident that adsorption of safranin-O increases with increase in initial dye concentration. This is because, when the safranin-O concentration increases in the solution, the active sites of the adsorbent will be surrounded by much more safranin-O molecule of it. Therefore, the percentage removal increases with increase in the initial concentration of the dye. The removal percentage increases very slightly with increase in temperature. This may be due to that, the number of binding sites increases with increase in temperature, hence enhancing the adsorption process.³⁸ A maximum dye removal (94.09%) resulted at temperature (29.59 °C) and initial dye concentration (38.65 mg L⁻¹).

3.5.3. Effect of temperature and adsorbent dose on adsorption process. The interactive effect of temperature and adsorbent dose on adsorption at constant pH (7) and initial dye concentration (30 mg L⁻¹) is shown in Fig. 6c. It may be noted that the dye adsorption decreases with increase in temperature and increase with increasing absorbent dose within the experimental range. The reason of this observation is thought to be the fact that, an increase in adsorbent dose in a solution of a constant dye concentration, increase the surface area and hence more availability of active site to safranin-O molecules caused an enhancement in dye removal. A maximum dye uptake (93.78%) was observed at temperature (31.61 °C) and adsorbent dose (0.66 mg L⁻¹).

3.5.4. Effect of pH and initial dye concentration on adsorption process. Fig. 7a represents the combined effect of pH and initial dye concentration on removal of safranin-O dye onto activated red mud at constant adsorbent dose (0.55 mg L⁻¹) and temperature (35 °C). It is evident that the percentage removal of dye increases with the increase in both pH and the dye concentration, as in the cases discussed earlier. The maximum removal of dye was observed at higher pH, this may be due to the pH_zpc of the adsorbent. A maximum dye removal (94.14%) was determined at pH (8.35) and initial dye concentration (37.03 mg L⁻¹).

3.5.5. Effect of pH and adsorbent dose on adsorption process. Fig. 7b depicted the combined effects of the adsorbent dosage and the initial pH for the dye adsorption by the activated red mud at constant temperature (35 °C) and initial concentration (30 mg L⁻¹). The percentage removal of safranin-O increases with increase in both adsorbent dose and the solution pH. As already mentioned above, at higher pH of the solution, a great number of OH⁻ in the experiment solution. On the other hand, at lower pH, there are a considerable number of OH⁻ in the aqueous medium. Therefore, the surface of the adsorbent can provide more negative charges in basic medium rather than acid medium, and since the dye is cationic, removing dye from the basic solution will be easier. In addition, the change of adsorbent dose increase the dye removal percentage even in higher pH. A maximum dye removal (94.01%) was obtained at pH (8.22) and adsorbent dose (0.68 mg g⁻¹).

3.5.6. Effect of initial dye concentration and adsorbent dose on adsorption process. The combined effect of initial dye concentration and adsorbent dose at constant temperature (35 °C) and pH (7) is shown in Fig. 7c. It is clear from the figure that, both initial dye concentration and adsorbent dose shows positive effects on percentage removal of safranin-O from aqueous solution. When initial dye concentration increase, the removal of dye increases with increase in adsorbent dose. This trend may be due to that, at lower dye concentrations and lower adsorbent dose, there are not sufficient dye molecules in the solution to adsorb on lees available binding sites of the adsorbent. Thus an increase in the adsorbent dose and dye concentration, the adsorption will be relatively higher due to presence of enough dye molecules in the solution to be adsorbed by more active binding sites provided by the adsorbent. At initial dye concentration (36.49 mg L⁻¹) and adsorbent dose (0.60 mg g⁻¹), a maximum dye uptake of 93.95% was reached.

3.6. Conformation of the optimal conditions

The percentage removal and the operation conditions for maximum percentage removal of safranin-O from aqueous solution was calculated from the second-order equation obtained from the experimental data. The first-order partial differential equation obtained from eqn (5) for X_i are:

The second-order differential equations are:

The value of X_i (X₁, X₂, X₃ and X₄) could be obtained after solving eqn (6)–(9) when ∂Y/∂X_i = 0, which gives the maximum value of Y (% removal). The solution of the above eqn (6)–(9) was found to be X₁ = 0.117, X₂ = −0.212, X₃ = 0.254 and X₄ = 0.382. Then the values were converted to actual values (uncoded) of X₁ (adsorbent dose) = 0.62 mg g⁻¹, X₂ (temperature) = 29.06 °C, X₃ (pH) = 8.3 and X₄ (initial concentration) = 37.3 mg L⁻¹ according to Table 2. At these optimum condition, the maximum predicted percentage removal of safranin-O dye was 94.5 ± 01% shown in Fig. 8.

Fig. 8 Optimum removal efficiency (contour plot obtained from RSM optimization).

3.7. Model verification

Five confirmation runs with a duplicate set were performed at the selected optimum conditions, in order to confirm the validity of the RSM model. The conditions are listed in Table 6. The dye removal conditions for the first run was done in new operation conditions within the range of the levels that were not conducted. The next two confirmation experiments are taken from the Table 3, while the last two runs were the operation conditions for maximum percentage removal of safranin-O dye. The error percentage among the actual and calculated values ranges from 0.38% to 10.11%. Therefore, it could be concluded that the second-order polynomial regression equation was able to predict the percentage removal of safranin-O dye from aqueous solution accurately.

Table 6 Results of confirmation experiments

(X1)	(X2)	(X3)	(X4)	% Removal
				Observed	Predicted
0.55	60	2	50	77.62	77.58
1	60	12	50	77.24	75.46
0.1	10	12	10	67.51	67.47
0.6	29	8.3	37	94.02	94.49
0.6	29	8.3	37	93.89	94.49

---

3.8. Adsorption isotherm models

In order to understand the feasibility of adsorption process, interaction between adsorbate–adsorbent and to know the adsorption capacity of the activated red mud, the equilibrium data were correlated with the theoretical Langmuir (eqn (14)), Freundlich (eqn (15)) and Temkin (eqn (16)) isotherm models.

q_e = kC_e^1/n (15)

where q_e is the amount of safranin-O adsorbed per unit mass of the adsorbent (mg g⁻¹), q_m is the maximum adsorption capacity (mg g⁻¹), b is the Langmuir constant related to the free energy of adsorption (L mg⁻¹), k is the Freundlich adsorption constant, n is constant parameter, B_T is the Temkin isotherm constant and K_T is the equilibrium binding constant (L mg⁻¹).

For the study of adsorption models, data were plotted as an adsorption capacity (q_e) as a function of equilibrium concentration (C_e) as shown in Fig. 9. The analysis of the different models (Langmuir, Freundlich, and Temkin model) were performed by means of a nonlinear fitting procedure. The Langmuir isotherm model was best fitted with the experimental data with the high R² value (0.994). The maximum adsorption capacity q_m and Langmuir constant b were found to be 9.768 mg g⁻¹ and 0.197 L mg⁻¹ respectively, shown in Table 7.

Fig. 9 Adsorption isotherm for safranin-O onto CO₂ neutralized activated red mud at optimum condition (obtained from RSM optimization).

Table 7 Langmuir, Freundlich and Temkin isotherm parameters at optimum condition

Isotherm model parameters
Langmuir isotherm		Freundlich isotherm		Temkin isotherm
qm	9.7680	1/n	0.1897	BT	0.6693
b	0.1972	k	4.3063	KT	8.3078
R^2	0.9996	R^2	0.9977	R^2	0.9987

---

4. Conclusions

In this study, a central composite design with the RSM was successfully applied to establish the optimum condition for the removal of safranin-O from aqueous solution by activated red mud. A quadratic polynomial equation was developed by RSM to predict the percentage removal of safranin-O and describe the relationship between response and variables. The model fitted the experimental data well, with a coefficient of determination, R² of 0.9549 and an adj-R² of 0.9312. Analysis of the model variance (ANOVA) showed that there was a high coefficient between adsorbent dose with initial concentration and reaction temperature with initial concentration of safranin-O dye. The optimum operating condition for adsorption of safranin-O dye were obtained as adsorbent dose of 0.62 g, temperature of 29.06 °C, pH of 8.3 and initial safranin-O concentration of 37.3 mg L⁻¹. The optimal response obtained from the RSM was 94.5 ± 0.1%, which is very close to the experimental value of 93.38%. The isotherm study shows that, the adsorption data was best fitted to Langmuir model with high correlation value of 0.994 and adsorption capacity of 9.7680 mg g⁻¹. It means that 1 g of red mud remove 9.7680 mg of safranin-O from 37.3 ppm safranin-O containing wastewater. At optimum condition (pH = 8.3, and temperature = 29.06), 1 tons of CO₂ neutralized activated red mud can remove 9.768 kg of safranin-O from 261 876 litres of standard waste water containing 37.3 ppm safranin-O. Therefore, it might be concluded that the activated red mud neutralized by CO₂ could be used as a promising adsorbent for removal of safranin-O dye from aqueous systems.

Acknowledgements

The authors are thankful to Vedanta Pvt. Ltd, Lanjigarh, Odisha, India, for funding the research project. The authors are also thankful to Director S. K. Sarangi, National Institute of Technology, Rourkela for providing the research facilities.

References

1. V. K. Garg, R. Kumar and R. Gupta, Dyes Pigm., 2004, 62, 1–10.
2. B. Royer, N. F. Cardoso, E. C. Lima, T. R. Macedo and C. Airoldi, J. Hazard. Mater., 2010, 181, 366–374.
3. Y. Aldegs, M. Elbarghouthi, a. Elsheikh and G. Walker, Dyes Pigm., 2008, 77, 16–23.
4. G. Crini, Bioresour. Technol., 2006, 97, 1061–1085.
5. M. Koch, A. Yediler, D. Lienert, G. Insel and A. Kettrup, Chemosphere, 2002, 46, 109–113.
6. P. K. Malik and S. K. Saha, Sep. Purif. Technol., 2003, 31, 241–250.
7. S. K. Nataraj, K. M. Hosamani and T. M. Aminabhavi, Desalination, 2009, 249, 12–17.
8. T. Panswad and S. Wongchaisuwan, Water Sci. Technol., 1986, 18, 139–144.
9. M. K. Purkait, S. DasGupta and S. De, Sep. Purif. Technol., 2004, 37, 81–92.
10. S. Seshadri and P. L. Bishop, Waste Manag., 1994, 15, 127–137.
11. N. Thinakaran, P. Baskaralingam, M. Pulikesi, P. Panneerselvam and S. Sivanesan, J. Hazard. Mater., 2008, 151, 316–322.
12. F. C. Wu and R. L. Tseng, J. Hazard. Mater., 2008, 152, 1256–1267.
13. M. Asgher and H. N. Bhatti, Ecol. Eng., 2012, 38, 79–85.
14. H. N. Bhatti, N. Akhtar and N. Saleem, Arabian J. Sci. Eng., 2012, 37, 9–18.
15. M. El Haddad, R. Slimani, R. Mamouni, M. R. Laamari, S. Rafqah and S. Lazar, J. Taiwan Inst. Chem. Eng., 2013, 44, 13–18.
16. I. Haq, H. N. Bhatti and M. Asgher, Can. J. Chem. Eng., 2011, 89, 593–600.
17. M. R. Malekbala, S. Hosseini, S. Kazemi Yazdi, S. Masoudi Soltani and M. R. Malekbala, Chem. Eng. Res. Des., 2012, 90, 704–712.
18. E. A. Moawed and A. B. Abulkibash, J. Saudi Chem. Soc., 2012 DOI:10.1016/j.jscs.2012.10.011.
19. Y. Safa and H. N. Bhatti, Desalination, 2011, 272, 313–322.
20. G. Power, M. Gräfe and C. Klauber, Hydrometallurgy, 2011, 108, 33–45.
21. S. Rai, K. L. Wasewar, J. Mukhopadhyay, C. K. Yoo and H. Uslu, Arch. Environ. Sci., 2012, 6, 13–33.
22. R. C. Sahu, R. Patel and B. C. Ray, J. Hazard. Mater., 2010, 179, 1007–1013.
23. D. Dorner, M. Baroni, C. Besson, C. Hood, F. Kesicki and P. Olejarnik, Redrawing the Energy-Climate Map: World Energy Outlook Special Report, 2013.
24. M. El Haddad, A. Regti, R. Slimani and S. Lazar, J. Ind. Eng. Chem., 2014, 20, 717–724.
25. M. A. Mohammed, A. Ibrahim and A. Shitu, Int. J. Environ. Monit. Anal., 2014, 2, 128–133.
26. S. Chowdhury, R. Mishra, P. Kushwaha and P. Saha, Asia-Pac. J. Chem. Eng., 2012, 7, 236–249.
27. S. Chowdhury, R. Misra, P. Kushwaha and P. Das, Biorem. J., 2011, 15, 77–89.
28. S. Preethi, a. Sivasamy, S. Sivanesan, V. Ramamurthi and G. Swaminathan, Ind. Eng. Chem. Res., 2006, 45, 7627–7632.
29. M. Amini, H. Younesi, N. Bahramifar, A. A. Z. Lorestani, F. Ghorbani, A. Daneshi and M. Sharifzadeh, J. Hazard. Mater., 2008, 154, 694–702.
30. W. Ma, X. Song, Y. Pan, Z. Cheng, G. Xin, B. Wang and X. Wang, Chem. Eng. J., 2012, 193–194, 381–390.
31. K. Ravikumar, K. Pakshirajan, T. Swaminathan and K. Balu, Chem. Eng. J., 2005, 105, 131–138.
32. B. Das, N. K. Mondal, P. Roy and S. Chattoraj, Int. J. Environ. Pollut. Solutions, 2013, 2, 72–87.
33. S. N. Su, H. L. Nie, L. M. Zhu and T. X. Chen, Bioresour. Technol., 2009, 100, 2336–2340.
34. J. Fan, C. Yi, X. Lan and B. Yang, Org. Process Res. Dev., 2013, 17, 368–374.
35. M. Evans, Optimization of Manufacturing Processes: A Response Surface Approach, Carlton House Terrace, London, 2003.
36. A. Özcan, E. M. Öncü and A. S. Özcan, J. Hazard. Mater., 2006, 129, 244–252.
37. F. Deniz, Sci. World J., 2013, 2013, 1–8.
38. R. Han, J. Zhang, P. Han, Y. Wang, Z. Zhao and M. Tang, Chem. Eng. J., 2009, 145, 496–504.

---