Sage Journals: Discover world-class research

Abstract

In this study, natural clay, iron oxide, and iron powder were combined to develop low-cost iron mixed porous pellet adsorbent for arsenate and arsenite removal from aqueous solution in batch experiments. The augmented simplex centroid mixture design was applied to obtain the optimum proportion of each constituent. Higher correlation coefficient of the models (R²> 0.95), good distribution of residuals, and lower values of p value (<0.05) indicated that the method is suitable for determining the optimum mixture proportion. Extensive decrease of both arsenate and arsenite adsorption occurred in the alkaline condition (pH > 9). Kinetic and isotherm experimental data of both arsenate and arsenite were well described by the pseudo-second order and Sips models, respectively. The maximum adsorption capacity of arsenate and arsenite derived from Sips model were 13.33 and 19.06 mg/g, respectively. The separation and heterogeneity factors showed that both arsenate and arsenite were favorably adsorbed. Among coexisting anions, phosphate significantly showed negative effect on the adsorption of either arsenate or arsenite. The adsorbent could be effectively reused for several times after its regeneration and was considered as non-hazardous material after adsorption.

Keywords

Porous adsorbent augmented simplex-centroid mixture design adsorption kinetics isotherm

Introduction

Contaminated water causes sickness or death for millions of people, especially those living in developing countries. Toxic pollutant is one of the major sources of contamination and among of the most highly toxic and carcinogenic elements is arsenic (As). The United State Environmental Protection Agency (US EPA) considers it the number 1 toxin in its list of prioritized pollutants and the International Agency for Research on Cancer categorizes it as a Class I human carcinogen (Ng et al., 2003). Arsenic, either resulting from natural processes or anthropogenic activities, has polluted drinking water sources in more than 70 countries and has put more than 100 million people worldwide at health risk (Brammer and Ravenscroft, 2009; Sigdel et al., 2016). The health problem related to excessive and long-term exposure to arsenic contaminated water is commonly termed arsenicosis including skin issues, cancers (skin, kidney, bladder, and lung), diabetes, cardiovascular disease, adverse pregnancy outcomes, and neurological problems (Jezequel and Chu, 2006; Sharma et al., 2014). More than 2.5 million people can develop arsenicosis in the next 50 years (Glocheux et al., 2013). Due to its potential for adverse health consequences, the World Health Organization has decreased the maximum contaminant level of arsenic in drinking water from 50 ppb to 10 ppb (Masih et al., 2009).

Drinking arsenic contaminated water is a main pathway for the toxin to enter the human body since arsenic lacks taste, color, and odor (Mondal et al., 2013; Sharma et al., 2014). Normally, arsenite (As(III)) and arsenate (As(V)) are the common forms presented in natural water. As(V) predominates in oxygen-rich environments like surface water, whereas As(III) favors and predominates in reducing environments like groundwater and more toxic than As(V) (Genc-Fuhrman et al., 2005; Katsoyiannis and Zouboulis, 2002). As(III) is possibly oxidized to As(V) via various methods and generally a pretreatment is applied before any treatment. However, pretreatment can consume lots of time, add additional cost, and produce toxic by-products. So, removing both As(III) and As(V) from contaminated water is preferred. Among arsenic treatment techniques, adsorption is a more suitable and promising method due to its low-cost, high uptake performance, simple operation and maintenance, scale flexibility, and sludge-free generation (Masih et al., 2009; Sabbatini et al., 2009). Arsenic removal efficiency through adsorption significantly depends on the nature and characteristic of an adsorbent that can either be natural or synthetic materials such as commercial and synthetic activated carbons, agricultural products or by-products, industrial by-products or wastes, and metal oxides (Mohan and Pittman, 2007). Currently, developing a low-cost and effective adsorbent with natural materials has gained more attention because of the cost reduction and plausibility of a real application.

Natural clay is popularly used as an adsorbent for removing pollutants owing to its large surface area, abundance and availability, low-cost, recycled ability, and non-toxicity (Kumrić et al., 2013; Sdiri et al., 2011; Srinivasan, 2011). However, natural clay expresses low adsorption efficiency toward arsenic most probably due to negative charge on its surface. Hence, applying natural clay to support metal oxides to enhance the adsorption capacity is needed. Of the metal oxides, iron oxide and elemental iron are well known for high affinity and adsorption capacity to arsenic and widely used to remediate arsenic from water; however, application in powder forms of those metal oxides expresses a difficulty of liquid and solid separation after adsorption process and it is very expensive to synthesize for large scale application (Bhowmick et al., 2014; Dousova et al., 2009; Sigdel et al., 2016). Clay supported metal oxides has been proved to remove some pollutants through various studies such as Pb(II) and Zn(II) removal by ceramisite produced from bentonite, iron powder, and activated carbon (Yuan and Liu, 2013); As(V) adsorption on adsorbents made from clay, iron oxide, and starch (Chen et al., 2010); Cr(VI) and As(V) removal by zero-valent iron and iron oxide-coated sand adsorbent (Mak et al., 2011); As(III) and As(V) removal by montmorillonite-supported zero valent iron (Bhowmick et al., 2014); and As removal by iron mixed ceramic pellet (Shafiquzzam et al., 2013). However, application of a systematic method to determine optimal proportion of each material for more effective adsorbent to remove both As(III) and As(V) is still limited.

The present study mainly aimed to develop porous pellets produced from natural clay to support iron oxide and iron powder to be a low-cost and effective adsorbent for both As(III) and As(V) removal from water. The augmented simplex centroid mixture design was applied to optimize the proportion of each constituent for the high response adsorption efficiency. The new optimized porous pellet adsorbent was characterized and also investigated for As(III) and As(V) adsorption in term of kinetics, isotherms, effect of solution pH, effect of coexisting anions, regeneration and leaching test.

Materials and methods

Materials and chemical reagents

Natural clay used in this study was collected from Dankwian District, Nakhon Ratchasima 30000, Thailand. The clay was cleaned with deionized (DI) water to remove debris and large particles, dried at 104 ± 1℃ for 24 h, crushed and sieved to achieve the particle size passing through a 200-mesh (<75 µm) sieve. Iron oxide powder (Fe₂O₃ > 95%, Himedia, India) was supplied by Italmar chemical supply company (Thailand). Iron powder (Fe > 96%, Xuzhou First Chemical Co., Ltd.) was purchased from a local supply store. Rice bran powder (chemical composition: P₂O₅ 53.744 wt.%, K₂O 17.316 wt.%, SO₃ 12.828 wt.%, MgO 7.569 wt.%, SiO₂ 5.362 wt.%, and CaO 2.237 wt.%) was obtained from a local grinding mill. As(V) and As(III) stock solutions (100 mg/L) were prepared by dissolving appropriate amount of Na₂HAsO₄·7H₂O and NaAsO₂ (Sigma Aldrich), respectively. NaCl, NaNO₃, NaHCO₃, Na₂CO₃, Na₂SO₄, and NaH₂PO₄ were used to prepare for anion solutions of chloride (Cl⁻), nitrate ( ${NO}_{3}^{-}$ ), bicarbonate ( ${HCO}_{3}^{-}$ ), carbonate ( ${CO}_{3}^{2 -}$ ), sulfate ( ${SO}_{4}^{2 -}$ ), and phosphate ( ${PO}_{4}^{3 -}$ ), respectively. Moreover, NaOH and HCl were used for solution pH adjustment.

Adsorbent development

The porous pellet adsorbent was developed from mixing natural clay (NC), iron oxide powder (Fe₂O₃), iron powder (IP), and rice bran powder (RB) at a specific mass proportion with DI water. The ratios of NC:Fe₂O₃:IP were selected by the mixture design method and the mass proportion of RB was constantly kept at 15% of the total mass. The prepared powders were uniformly mixed with addition of water to form a paste. The paste was continually stirred and pressed into long cylindrical shape with 1–1.5 mm in diameter by a plastic syringe. The long pellet was cut into 1–2 mm in length, placed in clean crucibles, dried at 60℃ for 24 h, and further heated at 600℃ for 1 h in a muffle furnace to carbonize and produce pores. After cooling down, the finished product was kept in a dry and clean container for further experiments.

Mixture design and statistical analysis

Mixture design approach is a type of experimental design used for analyzing the relationships between mixture components and responses, and for determining the best proportions of mixture components to obtain a final product in accordance with the target (BahramParvar et al., 2015). The proportion of a component in the mixture must be between 0 and 1 and the sum of the proportion of all mixture components is equal to 1 (Rosales et al., 2015). In this study, a three component augmented simplex-centroid mixture design was used to formulate different mixtures of independent variables, i.e. NC (X₁, wt.%), Fe₂O₃ (X₂, wt.%), and IP (X₃, wt.%), and to assess the relationship of the mixtures to the responses, As(V) and As(III) adsorption efficiency (Y₁ and Y₂, respectively). For the aspect of economy and being a binder, the amount of NC should be used more than others, i.e. Fe₂O₃ and IP serving as active site providers to the adsorbent. Therefore, constraints on individual component were applied and selected to be 50% < X₁ < 90%, 5% < X₂ < 45%, and 5% < X₃ < 45% in the form of X₁ + X₂ + X₃ = 100%. Totally, 10 different mixture design points were obtained and illustrated in Figure 1. The general polynomial function of the mixture models is represented by the following equation (Scheffe, 1963):

Y_{i} = \sum_{i = 1}^{q} a_{i} X_{i} + \sum \sum_{i < j}^{q} a_{ij} X_{i} X_{j} + \sum \sum_{i < j < k}^{q} \sum a_{ijk} X_{i} X_{j} X_{k}

(1)

where

Y_{i}

represents the predicted responses; q is the number of components in the mixtures;

X_{i}, X_{j}, X_{k}

are the represented independent variables; and

a_{i}, a_{ij}, a_{ijk}

are the coefficient of regression models for the linear, binary, and ternary mixture systems, respectively. Minitab statistical software (version 16.0, Minitab Inc., State College, PA) was used for designing the mixtures, analyzing the data, fitting the models, and optimizing constituent proportions.

Figure 1.

The overall instinct design points based on the three-component augmented simplex-centroid design.

Batch adsorption study

Batch mode was employed to conduct adsorption experiments that were carried out at room temperature (25 ± 1℃) in acid washed 60 mL polyethylene bottles with an agitation speed of 150 rpm on a horizontal mechanical shaker (New Brunswick Scientific, Canada). For kinetic study, a series of adsorption between 10 g/L of adsorbent and 10 mg/L of either As(V) or As(III) solution (pH = 7 ± 0.1) was prepared and shaken for 0 to 72 h. While shaking, samples were taken at preset time intervals. Isotherm experiments were performed with As(V) or As(III) concentration ranging from 0.5 to 100 mg/L. Effect of solution pH was conducted by varying pH from 3 to 11. Effect of coexisting anions were investigated using 10 g/L of adsorbent and 10 mg/L As(V) or As(III) solution (pH = 7 ± 0.1) containing 0.1, 1, and 10 mM of Cl⁻, ${NO}_{3}^{-}$ , ${HCO}_{3}^{-}$ , ${CO}_{3}^{2 -}$ , ${SO}_{4}^{2 -}$ , and ${PO}_{4}^{3 -}$ . The samples were filtered through 0.22 µm syringe filters and the filtrates were acidified with 1% v/v concentrated HNO₃, kept at 4℃ and analyzed for As(V) or As(III) concentration within 24 h. The adsorption capacity and efficiency (Q_e (mg/g) and R(%), respectively) were determined by equations defined as (Chutia et al., 2009)

Q_{e} = \frac{(C_{o} - C_{e})}{M} \times V

(2)

R (%) = \frac{(C_{o} - C_{e})}{C_{o}} \times 100

(3)

where C_o and C_e are the initial and equilibrium As(V) or As(III) concentration, respectively (mg/L), V is the volume of As(V) or As(III) solution (L), and M is the mass of adsorbent (g).

Adsorbent regeneration and arsenic leaching test

In order to observe the potential reuse of the adsorbent, three consecutive regeneration cycles were conducted. The adsorbed adsorbent of 10 g/L was dispersed in 25 mL of 1 M NaOH and shaken for 24 h. The desorbed adsorbent was washed several times with DI water and then reused to remove 10 mg/L of As(V) or As(III). Leaching of either As(V) or As(III) from the adsorbed adsorbent was examined using the US EPA Toxicity Characteristic Leaching Procedure (TCLP). The adsorbed adsorbent was mixed with the TCLP extraction fluid (5.7 mL of glacial acetic acid added to 500 mL of DI water and diluted to 1 L, pH 2.91) at a ratio of 1:20 and shaken with a mechanical shaker for 18 ± 2 h. The samples were filtered with 0.22 µm syringe filters and the filtrates were acidified and analyzed for As (V) or As(III) concentration.

Analytical methods

As(V) or As(III) concentration was measured by inductively coupled plasma-optical emission spectrometry (ICP-OES) (Optima 8000DV, PerkinElmer) using a wavelength of 193.7 nm. Elemental composition analysis was carried out with Energy Dispersive X-Ray Fluorescence (ED-XRF, XGT-5200 X-ray Analytical Microscope, HORIBA Ltd., Japan). The surface area, total pore volume, and average pore diameter were obtained from the Brunauer–Emmett–Teller (BET) method using the BET analyzer (BELSORP Mini II, BEL Inc., Japan). X-ray diffraction (XRD) for mineralogical phases was performed using the Bruker XRD (D2 PHASER). The morphological features of adsorbent before and after adsorption were examined by a scanning electron microscope (SEM, JSM-6010LV, JEOL, Japan) and the Fourier transform infrared spectra (4000–400 cm⁻¹ at a resolution of 4 cm⁻¹) was recorded using a PerkinElmer spectrum GX spectrophotometer (FT-IR) instrument. The point of zero charge (pH_pzc) was evaluated by the equilibrium method (Su et al., 2011).

Results and discussion

Model fitting and analysis of variance (ANOVA)

The detail of the experimental mixture design points and the experimental results of the responses are presented in Table 1. Each response was measured into two sets, Measurement 1 (M1) and Measurement 2 (M2) to minimize variance and improve the simulation for obtaining a better model.

Table 1.

Experimental design points and obtained responses, R(%)-As(V) and R(%)-As(III) measured in two sets (M₁ and M₂), for the corresponding mixtures.

Mixture	Natural clay (X₁)	Iron oxide (X₂)	Iron powder (X₃)	R(%)-As(V) (Y₁)		R(%)-As(III) (Y₂)
Mixture	Natural clay (X₁)	Iron oxide (X₂)	Iron powder (X₃)	M1	M2	M1	M2
A1	0.9	0.05	0.05	32.990	33.710	15.400	15.293
A2	0.5	0.45	0.05	67.728	67.860	84.263	82.430
A3	0.5	0.05	0.45	93.020	93.173	93.067	92.133
A4	0.7	0.25	0.05	53.135	51.463	47.897	43.543
A5	0.7	0.05	0.25	85.990	83.100	85.800	84.983
A6	0.5	0.25	0.25	91.263	91.730	92.633	92.660
A7	0.633	0.183	0.183	65.213	67.190	65.400	68.473
A8	0.766	0.116	0.116	58.203	59.588	51.900	54.693
A9	0.566	0.316	0.116	72.770	72.980	81.037	81.147
A10	0.566	0.116	0.316	94.330	94.500	94.130	93.010

Models with linear, quadratic, and special cubic degrees were applied to predict the independent variables and the responses. Run by the Minitab software, statistical parameters such as standard deviation (S), predicted sum of squares (PRESS), and predicted regression coefficient (R²) were obtained for linear (S = 7.224, PRESS = 1333.0, R²= 0.8145), quadratic (S = 4.466, PRESS = 437.3, R²= 0.9391), and special cubic models (S = 3.221, PRESS = 249.9, R²= 0.9652) in case of the As(V) response, and for linear (S = 9.907, PRESS = 2424.9, R²= 0.7973), quadratic (S = 4.331, PRESS = 389.7, R²= 0.9674), and special cubic models (S = 3.831, PRESS = 358.5, R²= 0.9700) in case of the As(III) response. The model with lower standard deviation and predicted sum of squares and higher predicted regression coefficient is a better one to predict the response data (Abdullah and Chin, 2010; Rao and Baral, 2011). Thus, special cubic model was the most suitable to fit the experimental data of both responses.

The ANOVA for the responses was evaluated based on 95% confidence level and p value was used to determine the significance of each component or the combination effect (Table 2). The special cubic model containing linear, two-component and three-component interactions is statistically significant for both responses (p < 0.001). In case of As(V) response, all terms were significant (p < 0.05). The interactions of components: X₁X₂, X₁X₃, X₂X₃, and X₁X₂X₃ (p value of 0.027, < 0.001, 0.001, and 0.003, respectively) had a significant effect. The interaction of X₁X₃ provided the highest significant effect due to its lowest p value. In case of As(III) response, component interaction X₁X₂, X₁X₃, X₂X₃, and X₁X₂X₃ influenced the response with p value of 0.534, <0.001, 0.045, and 0.045, respectively. Thus, X₁X₂ showed an insignificant effect and X₁X₃ provided the most significant influence on the As(III) removal efficiency.

Table 2.

Analysis of variance for all the responses.

Response	Source	DF	Seq SS	Adj SS	Adj MS	F	p value
As(V)	Regression	6	7050.42	7050.416	1175.069	113.22	<0.001
	Linear	2	6298.05	218.571	109.285	10.53	0.002
	Quadratic	3	608.03	637.744	212.581	20.48	<0.001
	X₁X₂	1	8.67	64.158	64.158	6.18	0.027
	X₁X₃	1	508.00	534.229	534.229	51.48	<0.001
	X₂X₃	1	91.35	167.741	167.741	16.16	0.001
	Special cubic	1	144.34	144.336	144.336	13.91	0.003
	X₁X₂X₃	1	144.34	144.336	144.336	13.91	0.003
	Residual error	13	134.92	134.919	10.378
	Total	19	7185.34
As(III)	Regression	6	11774.9	11774.89	1962.482	133.71	<0.001
	Linear	2	10297.0	737.51	368.754	25.12	<0.001
	Quadratic	3	1406.0	1442.18	480.727	32.75	<0.001
	X₁X₂	1	57.1	5.99	5.993	0.41	0.534
	X₁X₃	1	1339.6	819.58	819.579	55.84	<0.001
	X₂X₃	1	9.3	76.58	76.578	5.22	0.040
	Special cubic	1	71.9	71.86	71.863	4.90	0.045
	X₁X₂X₃	1	71.9	71.86	71.863	4.90	0.045
	Residual error	13	190.8	190.80	14.677
	Total	19	11965.7

The polynomial equations of the selected models for both responses can be expressed as follows:

Y_{1} = {2 X}_{1} - {18 X}_{2} - 20 {2 X}_{3} + {264 X}_{1} X_{2} + {762 X}_{1} X_{3} + {2422 X}_{2} X_{3} - {4258 X}_{1} X_{2} X_{3} with R^{2} = 0.9652

(4)

Y_{2} = - {24 X}_{1} + {145 X}_{2} - {266 X}_{3} + {944 X}_{1} X_{3} + {1637 X}_{2} X_{3} - 300 {5 X}_{1} X_{2} X_{3} with R^{2} = 0.9700

(5)

Residual graphs

The residual values are used to construct normal probability plots for evaluating the normal distribution of the data. The data are considered as normal distribution when the plotted points are close to the fitted straight line (Rostamiyan et al., 2014). Apparently, the plotted points for both cases seem closer to the distribution line, indicating the data were normally distributed (Figure 2).

Figure 2.

Normal probability plots for (a) As(V) response and (b) As(III) response.

Contour plots of the responses

A contour plot shows a detail effect on a certain response value by various establishing interactions of the mixture components in a two-dimensional view. Figure 3 illustrates the mixture contour plots of As(V) and As(III) responses resulting from the effect of the component interactions. The different desirable values of the responses were obtained and varied by changing in color. The darker color of the regions represents higher adsorption efficiency of the responses. For both responses, the maximum removal efficiency occurred toward the edge of Fe₂O₃ and IP with IP as vertex. This implies that adding Fe₂O₃ and IP proportion into the mixture improved both As(V) and As(III) adsorption efficiency to certain level.

Figure 3.

The mixture contour plots of the responses: (a) As(V) and (b) As(III).

Mixture proportion optimization and model validation

Response optimization involves defining independent variable settings that collaboratively produce optimized responses, and its satisfactory is measured by the composite desirability (Rao and Baral, 2011). The desirability is scaled from 0.0 (undesirable) to 1.0 (very desirable) (Shahamirifard et al., 2016). In the present study, the response optimization was processed using Response Optimizer of Minitab that identifies a combination of independent variables (NC, Fe₂O₃, and IP) to jointly optimize adsorption efficiency of As(V) and As(III). Figure 4 presents the optimization plot of independent variables affecting the predicted As(V) and As(III) adsorption efficiency. Independent variable settings on the plot could be adjusted by moving the vertical red line to obtain more desirable predicted responses. To obtain high composite desirability with a high comparable adsorption efficiency of both As(V) (91.3217%) and As(III) (91.8716%) by using less amount of iron oxide or iron powder as possible, the ideal mixture proportion of the materials is 0.5215 of NC, 0.1922 of Fe₂O₃, and 0.2863 of IP. The predicted As(III) desirability, As(V) desirability, and composite desirability were found to be d = 0.93048, d = 0.91826, and d = 0.9243, respectively, which were close to 1, indicating that the selected mixture ratio has a significant positive effect on maximizing the As(V) and As(III) adsorption efficiency (Rao and Baral, 2011). The optimum components were used to develop the porous pellet adsorbent for the As(V) or As(III) adsorption to validate the model. The results indicated that the As(V) and As(III) removal efficiency were 91.40% and 92.57%, respectively, comparable enough to the predicted values, suggesting that the predicted models were valid and adequate.

Figure 4.

The mixture optimization plot for As(V) and As(III) responses.

Characterization of the developed adsorbent

The BET surface area, total pore volume, and mean pore diameter of the porous pellet adsorbent were 18.267 m²/g, 0.0916 cm³/g, and 20.063 nm, respectively. The mean pore diameter within 2 to 50 nm suggested that the adsorbent is mesoporous according to the pore classification of the International Union of Pure and Applied Chemistry (Kuila and Prasad, 2013). Regardless the presence of oxygen, elemental composition analyzed by XRF indicated that the main constituent elements of porous pellet adsorbent were Si (21.829%), Al (7.251%), and Fe (25.780%) (Table 3). Apparently, the amount of iron element of the porous adsorbent has dramatically increased approximately 7 times compared to that of natural clay. Figure 5 shows the XRD pattern of the porous pellet adsorbent. The mineral constituents mainly consisted of quartz and hematite, as well as illite-montmorillonite and iron. The main crystalline phases of quartz and hematite were contributed by natural clay and iron oxide, respectively.

Table 3.

Chemical composition (wt.%) of natural clay and porous pellet adsorbent analyzed by XRF.

Composition (wt.%)	Si	Al	Fe	K	Ca	Ti	O
Natural clay	31.254	13.366	3.696	1.232	0.373	0.624	49.835
Porous pellet	21.829	7.251	25.780	0.626	0.308	0.342	43.246

Figure 5.

XRD pattern of the iron mixed porous pellet adsorbent.

Figure 6 shows the surface morphological features of the adsorbent before and after adsorption. The porous pellet adsorbent has a cylinder shape with numerous pores surrounding the surface, developed by carbonization of rice bran (Figure 6(a)). The scattered rough surface structure was observed before adsorption (Figure 6(b)). This indicated the spreading of iron oxide and iron powders to form heterogeneity on the surface. After adsorption of As(V) and As(III) (Figure 6(c) and (d), respectively), similar morphology of the surface were observed. The surface became smoother, consisted of small pores and seemed to be layered by various sizes of flat sheet. This is probably due to the binding between arsenic and iron species on the layer-structure of clay.

Figure 6.

SEM images of (a) shape of porous pellet (×25), (b) surface of porous pellet before adsorption, and porous pellet after adsorption (c) As(V) and (d) As(III).

The FTIR spectra of the adsorbent before and after As(V) and As(III) adsorption showed that the major peak bands were observed in the wavenumber range from 400 to 1200 cm⁻¹ (Figure 7). The peak band of 1029.94 cm⁻¹ and 800.42 cm⁻¹ was assignable to the stretching vibration of Si–O–Si group and Si–O deformation (Nayak and Singh, 2007; Petala et al., 2013; Tandon et al., 2013). The band of 692.41 cm⁻¹ corresponds to the spectrum for iron oxide (Fe₂O₃) (Li et al., 2010). The peak bands of 524.62 and 437.82 cm⁻¹ could be assigned to the vibration of Si–O–Al and Fe–O groups, respectively (Chen et al., 2011; Li et al., 2016). After As(V) and As(III) adsorption, reduction of stretching the peak bands, increase of transmittance percent of corresponding bands, and no presence of new peak bands were observed. This could be reasoned by As(V) and As(III) in the solution mostly adsorbed on the surface of the adsorbent (Chen et al., 2011).

Figure 7.

FTIR spectrum of porous pellet before and after As(V) and As(III) adsorption.

Adsorption kinetics

Kinetic study illustrates the chemical rate of adsorption process between adsorbate and adsorbent by observing the speed of a chemical reaction until reaching its equilibrium in a particular amount of time (Sen Gupta and Bhattacharyya, 2011). Adsorption performance of As(III) and As(V) onto porous pellet adsorbent dependent on time along with the fitting kinetic models showed a similar trend, which an initial wide availability of active sites provides a rapid uptake first and the filling up of active sites slows down and stabilizes the adsorption process within 72 h (Figure 8). Kinetic experimental data were fitted to the pseudo-first-order and pseudo-second-order models defined as (Far et al., 2012; Salameh et al., 2010)

Pseudo - firstorder : Q_{t} = Q_{e} (1 - e^{- k_{1} t})

(6)

Pseudo - secondorder : Q_{t} = \frac{k_{2} Q_{e}^{2} t}{{1 + k}_{2} Q_{e} t}

(7)

where Q_t (mg/g) is the amount of As(V) or As(III) adsorbed at time t; Q_e (mg/g) is the amount of As(V) or As(III) adsorbed at equilibrium time; k₁ (h⁻¹) is the rate constant of the pseudo-first order model; k₂ (g/mg.h⁻¹) is the rate constant of the pseudo-second-order model.

Figure 8.

Time dependent for (a) As(III) and (b) As(V) adsorption onto the adsorbent.

The correlation coefficients (R²) of the pseudo-second-order model (As(V): R²= 0.9919; As(III): R²= 0.9901) were higher than those of the pseudo-first-order model (As(V): R²= 0.9706; As(III): R²= 0.9824) and the calculated equilibrium adsorption capacity (As(V): 0.987 mg/g; As(III): 0.977 mg/g) was comparable enough to the experimental adsorption capacity (As(III): 0.914 mg/g; As(V): 0.906 mg/g) (Table 4). Therefore, it is concluded that the adsorption kinetics for both As(V) and As(III) could be better described by the pseudo-second-order rate model. This implied that the both As(III) and As(V) adsorption onto the adsorbent could be a chemisorption (Qi et al., 2015). From the pseudo-second-order model, the initial adsorption rate (h) calculated by h = k₂ Q_e² for As(V) and As(III) were 0.134 mg/g.h⁻¹ and 0.121 mg/g.h⁻¹, respectively, indicating that the adsorbent removed As(III) faster than As(V).

Table 4.

Kinetic parameters of As(III) and As(V) uptake onto the adsorbent.

As	Q_e,exp (mg/g)	Pseudo-first order			Pseudo-second order
As	Q_e,exp (mg/g)	Q_e (mg/g)	k₁ (h⁻¹)	R²	Q_e (mg/g)	k₂ (g/mg.h⁻¹)	R²
As(III)	0.914	0.727	0.0526	0.9706	0.977	0.1408	0.9901
As(V)	0.906	0.972	0.0548	0.9824	0.987	0.1242	0.9919

Effect of solution pH

As(V) adsorption capacity gradually decreased with the pH range from 3 to 7 and significantly decreased when pH was more than 7 (Figure 9). For As(III) adsorption, the removal capacity was not remarkably changed over the pH range from 3 to 9. Both As(V) and As(III) adsorption dramatically declined as pH raised above 9. Arsenic speciation and the characteristic of adsorbent surface importantly influence the adsorption capacity over various solution pH. In water, As(V) mainly exists in the form of H₃AsO₄ at pH less than 2.2, $H_{2} {AsO}_{4}^{-}$ at pH between 2.2 and 6.98, ${HAsO}_{4}^{2 -}$ at pH between 6.98 and 11.5, and ${AsO}_{4}^{3 -}$ at pH above 11.5 (Chang et al., 2010). As(III) species occur as: H₃AsO₃ (pH < 9.2), $H_{2} {AsO}_{3}^{-}$ (9 < pH < 12), ${HAsO}_{3}^{2 -}$ (12 < pH < 13), and ${AsO}_{3}^{3 -}$ (pH > 13) (Mohan and Pittman, 2007). The surface of the adsorbent was more positively charged for pH < pH_pzc and predominated the negative charge for pH > pH_pzc (Chang et al., 2010). The porous pellet adsorbent exhibited the point of zero charge (pH_pzc) of 7.50. This indicated that the unfavorable electrostatic interaction or the electrical repulsion between the adsorbent and the adsorbate could result in low As(V) and As(III) adsorption capacity at pH > 9.

Figure 9.

Effect of solution pH onto As(III) and As(V) adsorption by the adsorbent.

Isotherm study

Isotherm study was conducted to determine the characteristic of the adsorption and the maximum adsorption capacity of the adsorbent. The isotherm models such as the Langmuir, Freundlich, and Sips models were applied to predict the equilibrium isotherm data. The isotherm models were expressed as follows (Foo and Hameed, 2010; Jung and Ahn, 2016):

Freundlich : Q_{e} = K_{F} C_{e}^{1 / n}

(8)

Langmuir : Q_{e} = \frac{Q_{max} K_{L} C_{e}}{1 + K_{L} C_{e}}

(9)

Sips : Q_{e} = \frac{Q_{m} K_{L - F} C_{e}^{1 / n}}{1 + K_{L - F} C_{e}^{1 / n}}

(10)

where C_e (mg/L) is the As(V) or As(III) concentration at equilibrium; Q_e (mg/g) is the amount of As(V) or As(III) adsorbed at equilibrium per unit mass of the adsorbent; Q_max (mg/g) and Q_m (mg/g) are the maximum adsorption capacity based on the Langmuir and Sips models, respectively; K_L (L/mg), K_F (mg^1-1/nL^1/n/g), and K_L-F (L/mg) are the Langmuir constant, Freundlich constant, and Langmuir–Freundlich constant, respectively. 1/n is the heterogeneity factor. Sips model or Langmuir–Freundlich model is a combination of Langmuir (assuming a monolayer adsorption onto a homogenous surface) and Freundlich (assuming a multilayer adsorption onto a heterogeneous surface) (Jung and Ahn, 2016; Ouadjenia-Marouf et al., 2013). Experimental isotherm data and the fitting models are illustrated in Figure 10.

Figure 10.

Isotherm study of As(III) and As(V) adsorption onto the adsorbent.

Isotherm parameters such as the maximum adsorption capacity, constants of the fitting models, and correlation coefficient (R²) were calculated from non-linear forms of isotherm models (Table 5). The correlation coefficient (R²) of the Sips model was higher than that of the Langmuir and Freundlich for both As(V) and As(III) adsorption. This implied that the isotherm experimental data fit well to the Langmuir–Freundlich or Sips model, suggesting that As(V) or As(III) adsorption occurred as a monolayer on heterogeneous surface of the adsorbent (Chen et al., 2012). Obtained from the Sips model, the saturated removal capacities of As(V) and As(III) were 13.33 and 19.06 mg/g, respectively.

Table 5.

Isotherm parameters of As(III) and As(V) uptake onto the adsorbent.

As	Freundlich			Langmuir			Sips (Langmuir–Freundlich)
As	1/n	K_F	R²	Q_max (mg/g)	K_L	R²	Q_m (mg/g)	K_L-F	1/n	R²
As(III)	0.446	1.963	0.989	8.944	0.196	0.978	19.06	0.110	0.579	0.992
As(V)	0.302	1.940	0.989	5.403	0.467	0.948	13.33	0.173	0.398	0.993

The characteristic of the adsorption can be determined through either the value of a dimensionless constant separation factor (R_L) or the Freundlich constant (1/n). The separation factor was calculated by R_L = 1/(1 + K_LC_o), where C_o (mg/L) is the initial As(V) or As(III) concentration. The adsorption process is irreversible (R_L = 0), favorable (0 < R_L < 1), linear (R_L = 1), and unfavorable (R_L > 1) (Khan and Khan, 2015). Similarly, the magnitude of the adsorption intensity, 1/n, can suggest the type of isotherm from the Freundlich model. The adsorption is favorable when 0 < 1/n < 1, irreversible when 1/n = 1 and unfavorable when 1/n > 1 (Fufa et al., 2014). The values of 1/n for As(III) and As(V) were 0.446 and 0.302, respectively. Plus, for the studied initial concentration range, the R_L values for As(III) and As(V) adsorption were in the range of 0.05–0.91 and 0.02–0.81 (within 0 to 1). This showed that the adsorbent expressed a favorable adsorption toward As(V) or As(III). Noticeably, higher initial arsenic concentration tends to lead the adsorption to an irreversible isotherm and the adsorption approaches a linear isotherm at lower initial arsenic concentration.

Effect of coexisting anions

A similar trend of the influence of inorganic anions introduced separately and jointly on As(III) and As(V) adsorption was observed (Figure 11). The presence of Cl⁻ over the studied concentration range showed no significant effect on the adsorption. The anions ${NO}_{3}^{-}$ and ${SO}_{4}^{2 -}$ at higher concentrations could decrease the efficiency to 82.76% and 80.46%, respectively, for As(III) and 78.31% and 77.64%, respectively, for As(V). Increasing the concentration of ${CO}_{3}^{2 -}$ and ${HCO}_{3}^{-}$ could reduce the adsorption efficiency to 69.01% and 35.08%, respectively, for As(III) and 64.24% and 31.12%, respectively, for As(V). It was also observed that the presence of ${PO}_{4}^{3 -}$ strongly inhibited the removal of As(III) and As(V). The As(III) and As(V) efficiency decreased to 73.9% and 71.33%, respectively, for 0.1 mM ${PO}_{4}^{3 -}$ and it could be sharply reduced to 13.35% and 10.41%, respectively, for 10 mM ${PO}_{4}^{3 -}$ . The decrease observed for the mixture (combining all anions in the same solution) and for the presence of ${PO}_{4}^{3 -}$ was slightly different, which could be due to the capacity of ${PO}_{4}^{3 -}$ to surpass the buffering abilities of the other anions (Fufa et al., 2014). The reduction of adsorption efficiency when introducing the coexisting anions could be caused by the competition for active sites on the surface of the adsorbent. The remarkably negative influence of ${PO}_{4}^{3 -}$ could be explained by the reason that phosphate has similar chemical properties to arsenic and easily attaches to iron species by forming an inner-sphere complex compared to arsenic (Hsu et al., 2008; Maliyekkal et al., 2009).

Figure 11.

Effect of coexisting anions on the removal of (a) As(III) and (b) As(V) by the adsorbent.

Adsorbent regeneration and arsenic leaching test

The ability of the adsorbent to be regenerated and reused for As(III) and As(V) adsorption is presented in Figure 12. Apparently, the adsorption efficiency of As(III) and As(V) gradually decreased when the number of regeneration cycle increases. However, after the third regeneration time, the adsorbent could still maintain its well adsorption performance by being able to remove 84.94% and 83.06% for As(III) and As(V), respectively. In addition, the adsorbent expressed insignificant loss or no physical crush of the particle size. Thus, the adsorbent could be reused effectively for several times through desorption of NaOH treatment.

Figure 12.

Reuse of the adsorbent for As(III) and As(V) removal.

The toxicity characterization leaching procedure was applied to categorize the adsorbed adsorbent as either hazardous or non-hazardous material to be disposed. The US EPA recommends the permissible limit of 5 mg/L for the arsenic concentration in the leachate. The result indicated that As(III) and As(V) concentrations in the leachate were 0.54 and 0.73 mg/L, respectively, which were much lower than that of the limit. Therefore, the spent adsorbent was non-hazardous material and could be safely disposed as a solid waste.

Conclusions

Widely available natural clay could be easily used to bind iron oxide and iron powder to be the effective adsorbent for efficiently removing arsenate and arsenite from water. The augmented simplex-centroid mixture design method was a successful and useful tool to optimize the proportion of the constituent materials of the adsorbent. In the batch adsorption, the optimized adsorbent expressed a favorable adsorption toward both arsenate and arsenite even though the certain extent reduction of the efficiency occurred in the presence of some anions, especially phosphate. The adsorbent maintained high adsorption efficiency after several times of regeneration, was easily separated from the aqueous solution, and was non-hazardous solid waste after adsorption. In addition, regarded the low cost and availability of raw materials, the adsorbent could be cost-effective for arsenic removal from water.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Center for Scientific and Technological Equipment, and School of Environmental Engineering, Institute of Engineering, Suranaree University of Technology, Nakhon Ratchasima 30000, Thailand.

References

Abdullah

Chin

(2010) Simplex-centroid mixture formulation for optimised composting of kitchen waste. Bioresource Technology 101: 8205–8210.

BahramParvar

Tehrani

Razavi

et al. (2015) Application of simplex-centroid mixture design to optimize stabilizer combinations for ice cream manufacture. Journal of Food Science and Technology 52: 1480–1488.

Bhowmick

Chakraborty

Mondal

et al. (2014) Montmorillonite-supported nanoscale zero-valent iron for removal of arsenic from aqueous solution: Kinetics and mechanism. Chemical Engineering Journal 243: 14–23.

Brammer

Ravenscroft

(2009) Arsenic in groundwater: a threat to sustainable agriculture in South and South-east Asia. Environment International 35: 647–654.

Chang

Lin

Ying

(2010) Preparation of iron-impregnated granular activated carbon for arsenic removal from drinking water. Journal of Hazardous Materials 184: 515–522.

Chen

Lei

Yang

et al. (2012) Characterization and modification of porous ceramic sorbent for arsenate removal. Colloids and Surfaces A: Physicochemical and Engineering Aspects 414: 393–399.

Chen

Zhang

Feng

et al. (2010) Application of simplex-centroid mixture design in developing and optimizing ceramic adsorbent for As (V) removal from water solution. Microporous and Mesoporous Materials 131: 115–121.

Chen

Jin

Chen

et al. (2011) Removal of methyl orange from aqueous solution using bentonite-supported nanoscale zero-valent iron. Journal of Colloid and Interface Science 363: 601–607.

Chutia

Kato

Kojima

et al. (2009) Arsenic adsorption from aqueous solution on synthetic zeolites. Journal of Hazardous Materials 162: 440–447.

10.

Dousova

Fuitova

Grygar

et al. (2009) Modified aluminosilicates as low-cost sorbents of As(III) from anoxic groundwater. Journal of Hazardous Materials 165: 134–140.

11.

Far

Souri

Heidari

et al. (2012) Evaluation of iron and manganese-coated pumice application for the removal of as(v) from aqueous solutions. Iranian Journal of Environmental Health Science & Engineering 9: 1–9.

12.

Foo

Hameed

(2010) Insights into the modeling of adsorption isotherm systems. Chemical Engineering Journal 156: 2–10.

13.

Fufa

Alemayehu

Lennartz

(2014) Sorptive removal of arsenate using termite mound. Journal of Environmental Management 132: 188–196.

14.

Genc-Fuhrman

Bregnhoj

McConchie

(2005) Arsenate removal from water using sand–red mud columns. Water Research 39: 2944–2954.

15.

Glocheux

Pasarín

Albadarin

et al. (2013) Removal of arsenic from groundwater by adsorption onto an acidified laterite by-product. Chemical Engineering Journal 228: 565–574.

16.

Hsu

Lin

Liao

et al. (2008) Evaluation of the multiple-ion competition in the adsorption of As (V) onto reclaimed iron-oxide coated sands by fractional factorial design. Chemosphere 72: 1049–1055.

17.

Jezequel

Chu

(2006) Removal of arsenate from aqueous solution by adsorption onto titanium dioxide nanoparticles. Journal of Environmental Science and Health Part A 41: 1519–1528.

18.

Jung

Ahn

(2016) Fabrication of porosity-enhanced MgO/biochar for removal of phosphate from aqueous solution: Application of a novel combined electrochemical modification method. Bioresource Technology 200: 1029–1032.

19.

Katsoyiannis

Zouboulis

(2002) Removal of arsenic from contaminated water sources by sorption onto iron-oxide-coated polymeric materials. Water Research 36: 5141–5155.

20.

Khan

(2015) Removal of basic dyes from aqueous solution by adsorption onto binary iron-manganese oxide coated kaolinite: Non-linear isotherm and kinetics modeling. Applied Clay Science 107: 70–77.

21.

Kuila

Prasad

(2013) Specific surface area and pore-size distribution in clays and shales. Geophysical Prospecting 61: 341–362.

22.

Kumrić

Đukić

Trtić-Petrović

et al. (2013) Simultaneous removal of divalent heavy metals from aqueous solutions using raw and mechanochemically treated interstratified montmorillonite/kaolinite clay. Industrial & Engineering Chemistry Research 52: 7930–7939.

23.

et al. (2010) Synthesis and characterization of organo-montmorillonite supported iron nanoparticles. Applied Clay Science 50: 330–336.

24.

Zhao

et al. (2016) Removal of nitrobenzene by immobilized nanoscale zero-valent iron: Effect of clay support and efficiency optimization. Applied Surface Science 370: 260–269.

25.

Mak

Rao

(2011) Zero-valent iron and iron oxide-coated sand as a combination for removal of co-present chromate and arsenate from groundwater with humic acid. Environmental Pollution 159: 377–382.

26.

Maliyekkal

Philip

Pradeep

(2009) As(III) removal from drinking water using manganese oxide-coated-alumina: performance evaluation and mechanistic details of surface binding. Chemical Engineering Journal 153: 101–107.

27.

Masih

Seida

Izumi

(2009) Arsenic removal from dilute solutions by high surface area mesoporous iron oxyhydroxide. Water, Air, & Soil Pollution: Focus 9: 203–211.

28.

Mohan

Pittman

Jr (2007) Arsenic removal from water/wastewater using adsorbents–A critical review. Journal of Hazardous Materials 142: 1–53.

29.

Mondal

Bhowmick

Chatterjee

et al. (2013) Remediation of inorganic arsenic in groundwater for safe water supply: a critical assessment of technological solutions. Chemosphere 92: 157–170.

30.

Nayak

Singh

(2007) Instrumental characterization of clay by XRF, XRD and FTIR. Bulletin of Materials Science 30: 235–238.

31.

Wang

Shraim

(2003) A global health problem caused by arsenic from natural sources. Chemosphere 52: 1353–1359.

32.

Ouadjenia-Marouf

Marouf

Schott

et al. (2013) Removal of Cu(II), Cd(II) and Cr(III) ions from aqueous solution by dam silt. Arabian Journal of Chemistry 6: 401–406.

33.

Petala

Dimos

Douvalis

et al. (2013) Nanoscale zero-valent iron supported on mesoporous silica: characterization and reactivity for Cr(VI) removal from aqueous solution. Journal of Hazardous Materials 261: 295–306.

34.

Zhang

(2015) Efficient removal of arsenic from water using a granular adsorbent: Fe-Mn binary oxide impregnated chitosan bead. Bioresource Technology 193: 243–249.

35.

Rao

Baral

(2011) Experimental design of mixture for the anaerobic co-digestion of sewage sludge. Chemical Engineering Journal 172: 977–986.

36.

Rosales

Ferreira

Sanromán

MÁ

et al. (2015) Enhanced selective metal adsorption on optimised agroforestry waste mixtures. Bioresource Technology 182: 41–49.

37.

Rostamiyan

Hamed

SalmanKhani

(2014) Optimization of mechanical properties of epoxy-based hybrid nanocomposite: Effect of using nano silica and high-impact polystyrene by mixture design approach. Materials & Design 56: 1068–1077.

38.

Sabbatini

Rossi

Thern

et al. (2009) Iron oxide adsorbers for arsenic removal: A low cost treatment for rural areas and mobile applications. Desalination 248: 184–192.

39.

Salameh

Al-Lagtah

Ahmad

MNM

et al. (2010) Kinetic and thermodynamic investigations on arsenic adsorption onto dolomitic sorbents. Chemical Engineering Journal 160: 440–446.

40.

Scheffe

(1963) The simplex-centroid design for experiments with mixtures. Journal of the Royal Statistical Society. Series B (Methodological) 25: 235–263.

41.

Sdiri

Higashi

Hatta

et al. (2011) Evaluating the adsorptive capacity of montmorillonitic and calcareous clays on the removal of several heavy metals in aqueous systems. Chemical Engineering Journal 172: 37–46.

42.

Sen Gupta

Bhattacharyya

(2011) Kinetics of adsorption of metal ions on inorganic materials: A review. Advances in Colloid and Interface Science 162: 39–58.

43.

Shafiquzzam

Hasan

Nakajima

(2013) Iron mixed ceramic pellet for arsenic removal from groundwater. Environmental Engineering Research 18: 163–168.

44.

Shahamirifard

SAR

Ghaedi

Rahimi

et al. (2016) Simultaneous extraction and preconcentration of Cu²⁺, Ni²⁺ and Zn²⁺ ions using Ag nanoparticle-loaded activated carbon: response surface methodology. Advanced Powder Technology 27: 426–435.

45.

Sharma

Tjell

Sloth

et al. (2014) Review of arsenic contamination, exposure through water and food and low cost mitigation options for rural areas. Applied Geochemistry 41: 11–33.

46.

Sigdel

Park

Kwak

et al. (2016) Arsenic removal from aqueous solutions by adsorption onto hydrous iron oxide-impregnated alginate beads. Journal of Industrial and Engineering Chemistry 35: 277–286.

47.

Srinivasan

(2011) Advances in application of natural clay and its composites in removal of biological, organic, and inorganic contaminants from drinking water. Advances in Materials Science and Engineering 2011: 1–16.

48.

Huang

Jin

et al. (2011) Synthesis, characterization and kinetic of a surfactant-modified bentonite used to remove As(III) and As(V) from aqueous solution. Journal of Hazardous Materials 185: 63–70.

49.

Tandon

Shukla

Singh

(2013) Removal of arsenic(III) from water with clay-supported zerovalent iron nanoparticles synthesized with the help of tea liquor. Industrial & Engineering Chemistry Research 52: 10052–10058.

50.

Yuan

Liu

(2013) Removal of Pb(II) and Zn(II) from aqueous solution by ceramisite prepared by sintering bentonite, iron powder and activated carbon. Chemical Engineering Journal 215: 432–439.

Development of low-cost iron mixed porous pellet adsorbent by mixture design approach and its application for arsenate and arsenite adsorption from water

Abstract

Keywords

Introduction

Materials and methods

Materials and chemical reagents

Adsorbent development

Mixture design and statistical analysis

Batch adsorption study

Adsorbent regeneration and arsenic leaching test

Analytical methods

Results and discussion

Model fitting and analysis of variance (ANOVA)

Residual graphs

Contour plots of the responses

Mixture proportion optimization and model validation

Characterization of the developed adsorbent

Adsorption kinetics

Effect of solution pH

Isotherm study

Effect of coexisting anions

Adsorbent regeneration and arsenic leaching test

Conclusions

Footnotes

Declaration of Conflicting Interests

Funding

References