Arsenate removal from aqueous solution using chitosan- coated bentonite, chitosan-coated kaolinite and chitosan- coated sand: parametric, isotherm and thermodynamic studies

In the present work, the removal efficiency of As(V) from aqueous solution using chitosan-coated bentonite (CCB), chitosan-coated kaolinite (CCK) and chitosan-coated sand (CCS) was evaluated. The chitosan-based adsorbents were characterized using scanning electron microscopy, Fouriertransform infrared spectroscopy, the Brunauer–Emmett–Teller method and thermogravimetric analysis. Kinetic studies revealed that As(V) uptake using CCB, CCK and CCS fitted well with the pseudo-second order equation (R 0.9847; RMSE 9.1833). Equilibrium data show good correlation with the Langmuir model (R 0.9753; RMSE 8.5123; SSE 16.2651) for all adsorbents, which implies monolayer coverage onto homogenous energy sites. The Langmuir adsorption capacity for As (V) at pH 7.0 was determined to be 67.11, 64.85, and 16.78 mg/g for CCB, CCK and CCS, respectively. Thermodynamic studies show that As(V) uptake is exothermic in nature using CCK and endothermic using CCB and CCS. Moreover, adsorption of As(V) was feasible and spontaneous for CCB and CCS at 298 to 328 K. Results show that CCB is the most effective adsorbent in the removal of As(V) from water due to its high surface area and large pore diameter. doi: 10.2166/wst.2018.339 om https://iwaponline.com/wst/article-pdf/78/3/676/482294/wst078030676.pdf 2019 Cybelle M. Futalan National Research Center for Disaster-Free and Safe Ocean City, Busan 49315, Republic of Korea Yu-Shen Huang Department of Environmental Engineering and Science, Chia-Nan University of Pharmacy and Science, Tainan 71710, Taiwan Jheng-Hong Chen Department of Resources Engineering, National Cheng Kung University, Tainan 71710, Taiwan Meng-Wei Wan (corresponding author) Department of Environmental Resources Management, Chia-Nan University of Pharmacy and Science, Tainan 71710, Taiwan E-mail: peterwan@mail.cnu.edu.tw


INTRODUCTION
The contamination of arsenic in natural waters has been well documented in many countries including Argentina, Bangladesh, China, Chile, India, Philippines, Thailand and Taiwan (Mandal & Suzuki ; Sharma & Sohn ; Hosono et al. ). The primary source of arsenic in groundwater has been attributed to natural processes like geochemical reactions, volcanic emissions and weathering of arsenic rocks and minerals (Kanel et al. ). Another important source of arsenic is via anthropogenic activities such as ceramic and glassware production, herbicide and pesticide manufacturing, petroleum refineries, metallurgical industry and tannery operation (Altundogan et al. ).
Arsenic, which is a metalloid, is considered to be the 20th most abundant trace element (Mandal & Suzuki ; Nidheesh & Anantha Singh ). In general, arsenic exists in four oxidation states including þ5 (arsenate), þ3 and Taiwan have set the maximum contaminant level of arsenic for drinking water to be 25 and 10 μg/L, respectively (Chen & Chung ; Saldaña-Robles et al. ).
Adsorption is one of the most popular technologies utilized in the removal of heavy metals from waste effluents due to its simplicity in operation, low cost, safe handling, effectiveness even at a low concentration of contaminant, and applicability as a small-scale household module or in a community plant (Siddiqui & Chaudhry ). Adsorbents such as magnetite/non-oxidative graphene (Yoon et al. ), iron-coated seaweeds (Vieira et al. ), Fe 3 O 4 -GO-LDH (Wu et al. ), Fe 3 O 4 -GO (Sheng et al. ), chitosan-coated bentonite (CCB) (Arida et al. ) and perilla leaf biochar (Niazi et al. ) have been previously reported regarding the treatment of arseniccontaminated waters.
The utilization of bentonite, chitosan, kaolinite and sand, which are natural and readily available materials, as inexpensive, eco-friendly adsorbents is considered to be a sustainable approach when utilized in environmental remediation technology. Chitosan is derived from the deacetylation of chitin, which is extracted from the shells of shrimps, krills, crayfish and crabs (Kumari et al. ). The shrimp and fish industry generate crustacean waste residue composed of heads, shells and tails, which is 40-50% by weight of the total solid waste (Ogawa et al. ). Clay materials are often produced by mining industries and major civil infrastructure construction projects as overburden or waste rock (Ocak ; Lu & Cai ). Bentonite is a smectite-type of clay characterized by its 2:1 layered aluminosilicate sheets (Zhang et al. ) while kaolinite, Si 4 Al 4 O 10 (OH) 8 , has a 1:1 layered structure (Bhattacharyya & Gupta ). On the other hand, silica sand is an abundant, cost-effective material with properties such as hardness and resistance to heat and chemicals (Ramakrishna et al. ; Sundararajan et al. ).
Several studies have examined the use of CCB, chitosancoated sand (CCS) and chitosan-coated kaolinite (CCK) in the removal of various contaminants from wastewaters due to better chemical and mechanical stability, improved surface area and enhanced porosity of the adsorbent hybrid (Wan et al. ; Dalida et al. ; Futalan et al. ; Calagui et al. ; Arida et al. ; Chen et al. ; Ligaray et al. ). Moreover, coating chitosan onto a support material reduces the volume of chitosan required while retaining the overall adsorption efficiency (Wan et al. ). The work of Arida et al. () determined that the uptake capacity at breakthrough of As(V) is 10.57 μg/g under fixed-bed conditions. However, a comparative study on adsorption efficiency using CCB, CCK and CCS in the removal of arsenic under static conditions has not yet been reported.
The present study aims at evaluating the capacity of CCB, CCK and CCS for adsorption of As(V) from aqueous solution. Adsorbents were characterized using scanning electron microscopy (SEM), thermogravimetric analysis (TGA), Fourier transform infrared spectroscopy (FT-IR) and the Brunauer-Emmett-Teller (BET) multipoint technique. Kinetic studies were performed to determine the rate-limiting step of the adsorption system using pseudofirst order, pseudo-second order and intraparticle diffusion equations. Thermodynamic and equilibrium studies were also carried out.

MATERIALS AND METHOD Chemicals and materials
Low molecular weight chitosan (75-85% deacetylation), bentonite and kaolinite were purchased from Sigma-Aldrich (USA). Potassium iodide and hydrochloric acid (37% fuming) were acquired from Merck (Germany) while sand was obtained from Aquatek (Taiwan). Vitamin C and disodium hydrogen arsenate 7-hydrate (Na 2 HAsO 4 •7H 2 O) were purchased from Fisher Chemical (UK) and PanReac Applichem (Japan), respectively. All reagents obtained are of analytical grade and used without further purification.

Synthesis of CCB, CCK and CCS
The preparative method utilized was similar to that of Wan et al. () with some modifications. Chitosan (5 g) was dissolved in 300 mL of 5% (v/v) HCl by stirring for 2 h. Then, 100 g support material was added into the mixture and was stirred for 3 h. Addition of 1 M NaOH into the mixture was performed in a drop-wise method until pH 7 was attained. Adsorbent beads were washed with deionized water, filtered and dried in an oven (Channel Oven, DV452) for 24 h at 65 C. Adsorbents with a particle size range of 0.21 to 0.50 mm were utilized in the experiments.

Characterization
The morphology of the adsorbents was observed using SEM (S-3000N Hitachi, Japan) with a voltage of 20 kV under a vacuum atmosphere of 1.33 × 10 À6 mBar. The surface area and porosity of the adsorbents were determined using a BET surface analyzer (Micromeritics ASAP 2010, USA) at 77 K under N 2 gas. TGA was carried out using a thermomechanical analyzer (Perkin Elmer Pyris TGA-4000, USA) with a heating rate of 10 C min À1 within the range of 30 to 800 C. The spectra of chitosan, clay materials, sand, CCB, CCK and CCS were determined using FT-IR (Nicolet 6700) in the range of 400 to 4,000 cm À1 .

Batch adsorption study
Adsorption tests were carried out where 1 g adsorbent and 30 mL As(V) solution were agitated at 50 rpm using a reciprocal shaker bath (YIH-DER-BT 350, Taiwan) at 25 C and pH 7.0 under varying initial concentration (50 to 1,000 μg/L) and contact time (0.5 to 24 h). Thereafter, the solution was filtered through a 0.45 μm cellulose filter and the filtrate was analyzed using an inductively coupled plasma optical spectroscopy (ICP-OES, Perkin Elmer Optima 2000 DV, USA).

Data analysis
The adsorption capacity at equilibrium, q e (mg/g) was computed using Equation (1): where V is the solution volume (mL), M is the adsorbent mass (g), C 0 and C e refers to the initial and equilibrium concentration (mg/L), respectively.
To determine the rate-determining step of the adsorption system, kinetic equations are applied. The pseudo-first order (Lagergren ), pseudo-second order (Ho & McKay ) and intraparticle diffusion (Weber & Morris ) equations are expressed as the following: where q t (mg/g) denotes the adsorption capacity at time t (min), k id (mg/g•min 0.5 ) denotes the rate constant of the intraparticle diffusion model, C refers to the thickness of the boundary layer (mg/g), and k 1 (1/min) and k 2 (g/ mg•min) refer to the kinetic rate constants of the pseudofirst and pseudo-second order equation, respectively. Equilibrium data are fitted using Langmuir, Freundlich and Dubinin-Radushkevich (D-R) isotherms, which are written as Equations (5) to (9): ln q e ¼ ln q DR À βε 2 (7) where q m refers to the maximum adsorption capacity at monolayer coverage (mg/g), K L is the Langmuir constant that refers to the heat of adsorption (L/mg), 1/n refers to the linearity of adsorption (L/mg), K F is the Freundlich adsorption capacity of the adsorbent (mg/g), q DR is the D-R monolayer adsorption capacity (mg/g), β refers to the D-R constant regarding the mean free energy of the adsorbent (mol 2 /kJ 2 ), E is the free energy of adsorption (kJ/ mol), ε refers to the Polanyi potential (mol/kJ), R is the empirical gas constant (8.314 J/mol•K), and T is the operating temperature (K). The Langmuir isotherm assumes a monolayer adsorption occurring onto binding sites with homogenous energy levels where no further adsorption takes place at occupied sites (Langmuir ). The Freundlich isotherm, which is based on an empirical equation, refers to the adsorption of a solid adsorbent with energetically heterogeneous surfaces (Freundlich ; Wan Ngah & Fatinathan ). Meanwhile, the D-R isotherm is derived from the adsorption of sub-critical vapors on energetically irregular surfaces (Dubinin & Radushkevich ). Linear and non-linear methods were used in evaluating the kinetic and equilibrium data. In Microsoft Excel, the Solver add in function was utilized in data analysis using non-linear equations. The linear form of the models was assessed using the linear regression method.

Error analysis
Error functions are used in order to evaluate the fit of the kinetic and isotherm equations with the experimental data.
The sum of squares error (SSE) and root means square error (RMSE) were computed using Equations (10) and (11) (Karri et al. ): jq i e, exp l À q i e,theor j (10) where n is the number of observations, q i e,expl refers to the measured concentration (μg/L) and q i e,theor stands for the predicted value of the model (μg/L).  adsorbents can be arranged in the following order: CCB > CCK > CCS. Moreover, CCS has a smoother exterior morphology with few ridges and wrinkles covering its surface, while CCB and CCK displayed aggregation with a coarse surface.

RESULTS AND DISCUSSION
Based on BET analysis, both chitosan and CCK were determined to be mesoporous materials while CCB and CCS exhibited macroporosity. Pure chitosan displayed a low surface area (2.12 m 2 /g), pore volume (0.032 cm 3 /g) and diameter (11.61 nm) while coating chitosan onto a support material resulted in a higher BET surface area, pore volume and average pore diameter for CCB (9.22 m 2 /g, 0.048 cm 3 /g, 80.54 nm) and CCK (5.52 m 2 /g, 0.035 cm 3 /g, 29.98 nm). Among the adsorbents, CCS showed the least surface area (0.47 m 2 /g) and pore volume (0.004 cm 3 /g) but has the largest average pore diameter (683.80 nm).
Figure 1(d) to 1(f) illustrate the TGA curves of CCB, CCK and CCS. In general, chitosan displayed a weight loss of 12% that occurred between 70 C to 97 C, which is attributed to loss of physically adsorbed water. The second and third decomposition stages occurred between 232 C to 550 C and were due to decomposition of the chitosan backbone such as D-glucosamine and N-acetylglucosamine (Soon et al. ). Then, chitosan was burnt out completely at 550 C.
For bentonite, 5% weight loss at <100 C was due to loss of physically adsorbed water molecules in the interlayer (Figure 1(d)). A second weight loss of 4% at 400 C to 700 C was attributed to the evaporation of water molecules contained within the crystal lattice (Belbachir & Makhoukhi ). Figure 1(e) illustrates two decomposition stages of kaolinite: 2% loss at 200 C to 470 C and 6% loss at 470 C to 620 C due to physically adsorbed water and dihydroxylation of kaolinite samples, respectively (Drweesh et al. ). Meanwhile, sand displayed only 2% weight loss starting at 80 C due to loss of physically adsorbed water (Figure 1(f)). The curves of CCB, CCK and CCS are a combination of degradation profiles of chitosan and the support materials. When compared with the support materials' profile, the weight loss observed were higher such as 10%, 15% and 2.5% for CCB, CCK and CCS, respectively. Based on the TGA curves, 4.7%, 4.8% and 2.1% of chitosan were coated onto bentonite, kaolinite and sand, respectively. Figure 2 illustrates several peaks that are attributed to chitosan: 3,411 cm À1 refers to the primary amine and hydroxyl groups stretching vibration, 2,808 cm À1 refers to the stretching of C-H in CH 2 groups, 1,414 cm À1 refers to

Kinetic study
As shown in Figure 3(a) to 3(c), results displayed a rapid initial uptake of As(V) within 2 h that could be attributed to the availability of numerous adsorption sites. At an initial As(V) concentration of 1,000 μg/L, adsorption capacity of 23.02, 17.35 and 16.92 mg/g was attained for CCB, CCK and CCS, respectively. After which, adsorption capacity was observed to increase slowly until equilibrium was attained in 4 h at 100 μg/L and 12 h at 500 and 1,000 μg/L. A low concentration would indicate that there is less amount of As(V) present in the solution, hence there is less competition and ions bind easily onto the adsorbent. At higher concentration, the effect of steric crowding becomes more significant since there would be shorter distances between As(V) ions present in the solution, where repulsive forces would be exerted between ions, which would contribute to the delay in the attainment of equilibrium. Based on Table 1 and 2, the high values of the coefficient of determination (R 2 ) for CCB (R 2 ! 0.9970), CCK (R 2 ! 0.9923) and CCS (R 2 ! 0.9847) derived from linear and non-linear equation indicate a good correlation between the experimental and predicted data generated by the pseudo-second order equation. Figure 3(d) to 3(f) display the goodness of fit of the experimental data with the linear plots generated by the pseudo-second order model. Moreover, the low error values of RMSE for CCB, CCK and CCS further validate that the pseudo-second order equation best describes the adsorption system. This implies that the rate-determining step is chemisorption, wherein a covalent bond is formed through sharing a pair of electrons between As(V) and the active sites of the adsorbents. It is observed that the non-linear equations provided lower RMSE error values and higher R 2 values in comparison to the linear equation.
Under all concentration ranges studied, the highest k 2 constant occurred with CCB followed by CCK and CCS. The high rate constant of CCB would imply rapid adsorption rates and high adsorption capacity. For all adsorbents, the kinetic rate constant, k 2 , was observed to decrease as the initial concentration was increased from 100 to 1,000 μg/L. This denotes that there is tighter competition between ions at high concentration that leads to a lower rate constant and decreased adsorption capacity.

Isotherm study
Based on Table 3, the high R 2 values (R 2 ! 0.9753) and low error values of SSE and RMSE imply that the Langmuir isotherm best describes the adsorption system. This denotes that uptake of As(V) by CCB, CCK and CCS occurs as a monolayer coverage onto binding sites with homogenous sorption energies. Results show that the parameters derived from linear and non-linear equations do not have the same values. The discrepancy in the values is attributed to the use of linear equations (Narayanan et al. ). However, the parameters displayed a similar trend whether linear or non-linear equations were utilized. The values of the Langmuir constant q m can be arranged in the order: CCB > CCK > CCS, where the high adsorption capacity of CCB can be attributed to its high surface area and large pore diameter. Hence, the preferential adsorption of As(V) is due to the higher number of binding sites available for CCB. Moreover, a larger pore diameter would mean accessibility of As(V) with a thermochemical radius of 0.248 nm into the pore network, which results in better adsorption rates (Chen et al. ). The Freundlich constant n would indicate whether the adsorption process is chemical (n < 1) or physical (n > 1) in nature (Veli & Alyuz ). For CCB, CCK and CCS, the n values were determined to be within the range of 1.225 to 2.303, which implies that the adsorption system is physical in nature. Figure 4(a) to 4(c) illustrate the Freundluch, Langmuir and D-R models plotted against the experimental data. A good correlation is observed between experimental and predicted data generated by the Langmuir model, which further validates that Langmuir is the most appropriate isotherm to describe the adsorption process in the present work.
Among the isotherm models applied, D-R provided the lowest R 2 values (R 2 0.9658) and high error values (RMSE! 8.5436; SSE ! 12.8545) that signifies unsatisfactory correlation between predicted and experimental data. The D-R constant E describes the change in free energy when a mole of adsorbate from solution diffuses onto the adsorbent surface (Chen et al. ). Its value provides information on whether the adsorption behavior is related to physical adsorption (E < 8 kJ/mol), ion-exchange mechanism (8 kJ/ mol < E < 16 kJ/mol) or chemisorption (E > 16 kJ/mol). The values obtained for CCB, CCK and CCS lie within the range of 0.017 to 0.143 kJ/mol, which indicates that physical adsorption predominates in the uptake of As(V). In Table 4, the As(V) adsorption capacity of CCB, CCK and CCS was compared with various other adsorbents reported in the literature. Among the adsorbents, Fe 3 O 4 -GO-LDH provided the highest adsorption capacity followed by CCB and CCK. It is essential to emphasize that CCB and CCK are derived from low cost and readily available materials when compared to Fe 3 O 4 -GO-LDH, where the latter would require a more complicated fabrication procedure. Results indicate that CCB and CCK can be applied as low-cost adsorbents in removing As(V) from waste effluents under neutral conditions.

Thermodynamic study
The effect of temperature on the As(V) adsorption by CCB, CCK and CCS under varying temperature is shown in Figure 4(d). Results show that As(V) uptake increased from 10.81 to 11.72 mg/g as temperature was increased from 298 to 328 K for CCK while adsorption of As(V) decreased from 15.18 to 12.83 mg/g for CCB and from 8.75 to 4.98 mg/g for CCS with higher temperature.
Thermodynamic parameters such as standard Gibbs free energy (ΔG , kJ/mol), enthalpy (ΔH , kJ/mol) and entropy (ΔS , kJ/mol•K) were calculated using Equations (12) to (14) (Liu ): log where γ e is the activity coefficient at equilibrium, z is the charge of As(V) and I e is the ionic strength of As(V) at equilibrium. Equation (12) was utilized based on As(V) being a charged species under high concentration, where γ e has been taken into account. Based on Table 5, the negative values of ΔG 0 of CCB and CCS from 298 to 328 K indicate that the adsorption of As(V) proceeds spontaneously. Meanwhile, non-spontaneous adsorption occurs at 298 to 328 K for CCK. A more negative value of ΔG would favor adsorption. Since the value of ΔG for CCB yields to be the most negative, adsorption using CCB was favored over CCK and CCS. For CCB and CCS, the values of ΔG 0 become more negative with increase in temperature, implying that adsorption is more feasible at higher temperatures. This implies that adsorption was more feasible at 328 K due increase in the diffusion rates of the As(V). On the other hand, ΔG 0 values of CCK were observed to become more positive at higher temperature. An increase in temperature could indicate less stable bonds between the binding sites and As(V), which led to decreased adsorption capacity (Yoon et al. ). In all adsorbents, the positive ΔS 0 values imply that there is an increase in randomness after adsorption of As(V) onto the adsorbent surface. Meanwhile, the negative value of ΔH 0 indicates that As(V) uptake onto CCK is exothermic in nature while As(V) uptake using CCB and CCS is an endothermic process.

CONCLUSIONS
In the present study, the viability of utilizing CCB, CCK and CCS in the uptake of As(V) from aqueous solution was evaluated. Equilibrium data of the three adsorbents were fitted using the linear and non-linear isotherm models. Results show that the data correlated well with the Langmiur isotherm, which indicates that As(V) adsorption occurs at a monolayer coverage onto binding sites with homogenous energy levels. Based on both linear and non-linear methods, low values of error functions (RMSE 8.5123; SSE 16.2651) with high R 2 values (!0.9753) were attained for the Langmuir isotherm. At pH 7.0 and 298 K, the maximum adsorption capacity derived from the Langmuir model was determined to be 67.11, 64.85, and 16.78 mg/g for CCB,  CCK and CCS, respectively. In addition, the obtained values are comparable or higher than the values of various adsorbents reported in the literature. CCB was shown to have the highest adsorption capacity for As(V) over CCK and CCS, which suggests easily accessible and higher number of available binding sites due to its high surface area and large pore diameter. The E values (0.017 to 0.143 kJ/mol) derived from the D-R isotherm imply that the adsorption process is governed by physical adsorption. High R 2 values (!0.9847) and low error function values (RMSE 9.1833) indicate that the pseudo-second order equation best describes the kinetic data of As(V) adsorption. This denotes that chemisorption is the rate-limiting step of the adsorption system. Adsorption of As(V) results in an increase in the randomness at the solution-solid interface. Moreover, adsorption of As(V) by CCK is exothermic in nature and endothermic for CCB and CCS. This study demonstrated the feasibility of using CCB and CCK for the removal of As(V) from water.