Journal of Environmental Chemical Engineering 7 (2019) 102901 Contents lists available at ScienceDirect Journal of Environmental Chemical Engineering journal homepage: www.elsevier.com/locate/jece Synthesis and adsorption behavior of mesoporous alumina and Fe-doped T alumina for the removal of dominant arsenic species in contaminated waters ⁎ N. Inchaurrondoa, , C. di Lucaa, F. Morib, A. Pintarc, G. Žerjavc, M. Valiented, C. Paletd a Department of Chemical Engineering, Institute of Materials Science and Technology (INTEMA), University of Mar del Plata and National Research Council (CONICET), Av. J. B. Justo 4302, 7600, Mar del Plata, Argentina b FIQ-UNL, INCAPE-CONICET, Santiago del Estero, 2829, S3000AOM, Santa Fe, Argentina c Department for Environmental Sciences and Engineering, National Institute of Chemistry, Hajdrihova 19, SI-1001, Ljubljana, Slovenia d Centre Grup de Tècniques de Separació en Química, Departament of Chemistry, Universitat Autònoma de Barcelona, 08193, Barcelona, Catalunya, Spain A R T I C LE I N FO A B S T R A C T Keywords: Ordered mesoporous Al2O3 and Fe-Al2O3 materials were synthesized at room temperature by an easy and en- Mesoporous alumina vironmentally friendly self-assembly sol-gel route, to be tested for arsenic removal. Fe doped alumina Solid samples were thoroughly characterized by several techniques. Synthetized and commercial alumina As(III) and As(V) removal samples were evaluated as adsorbents for the removal of dominant arsenic species under a wide pH range Adsorption mechanism (3.6–11.5). The mesoporous alumina showed higher adsorption capacity (90 mg/g As(V), pHeq 4) than com- mercial alumina (54 mg/g As(V), pHeq 4), due to its amorphous structure, uniform accessible mesopores and higher surface acidity. The Fe bearing material exhibited strong As affinity. As(III) adsorption resulted much lower than for arsenate (maximum uptake of 16 mg/g, at pH 8), since As (III)-adsorbent interaction is only based on weak Van der Waals force. Arsenic isotherms adjusted well to the Freundlich model and more accurately to the three parameters Sip´s model. The kinetics results fitted the Elovich model. As pH increased, adsorption capacity decreased due to the reduction of electrostatic interactions. Under alkaline conditions arsenic desorption was achieved, although the stability of the material was compromised. The presence of several interfering ions was evaluated. Phosphate ions showed the highest interference. The use of a tap water matrix increased As(V) adsorption, encouraging the use of these materials in the treatment of real polluted waters. 1. Introduction groundwater [4]. In Argentina, arsenic contamination affects mainly urban and rural Arsenic contaminated waters are a matter of great concern in sev- poor populations, not connected to drinking water networks [4]. Due to eral areas of the world. The chronic exposure to this pollutant can lead its high toxicity, the World Health Organization (WHO) recommends a to adverse health effects such as skin lesions and lung, bladder, and limit of 10 ppb of arsenic in water for human consumption [1]. How- kidney cancer [1]. ever, in Argentina, the regions with the highest arsenic content in water Arsenic is mobilized by natural weathering reactions, biological (Chaco-Pampean plain, Puna and Cuyo), show values between 4 and activity, geochemical reactions, volcanic emissions and some anthro- 5.3 mg/L (mainly As(V)). Also, a high arsenic content of 14.97 mg/L pogenic activities (use of fertilizers, pesticides, glass, ceramics, mining/ was reported in the province of Santiago del Estero [4]. In some places, smelting operations, petroleum refineries) [2,3]. Arsenic appears as 99% of groundwater exceeds the guideline of 10 ppb [4], therefore it is inorganic as well as organic species; however, in an aqueous environ- imperative to optimize an affordable and simple operation arsenic re- ment, inorganic arsenic species (arsenite, As(III), and arsenate, As(V)) moval process. are more abundant than organic ones. Arsenate dominates at high redox Various technologies have been applied to solve this problem, such potentials and oxygen-rich aerobic environments, is less toxic than ar- as co-precipitation, alum or iron coagulation, membrane filtration, senite, but more abundant and mobile in natural surface waters; bioremediation or adsorption [5]. Among these alternatives, the ad- whereas, arsenite is found mostly in anaerobic environments such as sorption technology is very promising because of its simple operation, ⁎ Corresponding author. E-mail address: ninchaurrondo@gmail.com (N. Inchaurrondo). https://doi.org/10.1016/j.jece.2019.102901 Received 11 September 2018; Received in revised form 27 November 2018; Accepted 10 January 2019 Available online 11 January 2019 2213-3437/ © 2019 Elsevier Ltd. All rights reserved. N. Inchaurrondo et al. Journal of Environmental Chemical Engineering 7 (2019) 102901 Nomenclature pH0 Initial pH value pHeq Equilibrium pH value C0 Aqueous concentration of arsenic at initial time (mg/L) PZC pH value of point of zero charge CE Aqueous concentration of arsenic at equilibrium (mg/L) qe Solid phase concentration of arsenic at equilibrium (mg/g) dpore Average pore diameter (nm) qt Solid phase concentration of arsenic at any time (mg/g) k1 Pseudo-first order rate constant (min−1) Qa Maximum adsorption capacity per gram of adsorbent, k2 Pseudo-second order rate constant (g mg−1 min−1) Sip´s model (mg/g) α Elovich initial adsorption rate (g mg−1 min-2) Q’a Maximum adsorption capacity per square meter of ad- β Elovich desorption coefficient (mg g−1 min−1) sorbent, Sip´s model (mg/m2) kp Intra-particle diffusion model rate constant (mg Q Em Maximum adsorption capacity per gram of adsorbent (mg/ g−1 min−1) g) C Intra-particle diffusion model constant, related to thick- Q’Em Maximum adsorption capacity per square meter of ad- ness of boundary layer (mol2 J−2) sorbent (mg/m2) KRD DR constant related to sorption energy (mol2 J−2) QE Adsorption capacity per gram of adsorbent (mg/g) A Temkin equilibrium constant (l mg−1) Q’E Adsorption capacity per square meter of adsorbent (mg/ B Temkin variation of adsorption energy (J/mol) m2) Ka Sip’s adsorption affinity constant, where i stands for the Qm Maximum saturated monolayer adsorption capacity per pHeq value (L/mg) gram of adsorbent (mg/g) KF Freundlich adsorption coefficientper gram of adsorbent Q’m Maximum saturated monolayer adsorption capacity per (mgn−1 Ln/g) square meter of adsorbent (mg/m2) K’F Freundlich adsorption coefficientper square meter of ad- QRD Maximum adsorption capacity DR model (mg/g) sorbent (mgn−1 Ln/m2) RT Room temperature KL Langmuir adsorbent-adsorbate affinity constant (L/mg) SBET Specific surface area (m2/g) n Freundlich isotherm exponent V Volume of the arsenic solution (L) u Sip’s model exponent Vpore Pore volume (cm3/g) Table 1 Results presented by other authors employing alumina or iron oxides for arsenic removal. Adsorbent pHeq Surface area T Capacity (mg/g) Reference (m2/g) (ºC) As(V) As(III) IOCD-20,11 (diatomite/Fe) 5.5 93.0 30 8.60 – [3] Fe2O3 6 – RT 4.60 1.25 [9] Fe3O4 6 – RT 6.70 8.20 [9] Hematite 7 4.36 25 0.900 – [10] Goethite 7 58.2 25 7.60 – [10] Siderite 7 39.7 25 8.10 – [10] Ferric impregnated volcanic ash 7 147 20 5.30 – [11] Hematite 6 1.66 25 3.30 – [12] Goethite 6 11.6 25 1.40 – [12] Zero valent iron 6 0.200 25 30.1 – [12] Magnetite 6 1.60 25 7.10 – [12] Synthetic Fe-Al mixed oxides 7 131 30 54.5 – [13] Mesoporous γ-Al2O3 6.5-7 110 RT 19.8 – [14] Mixed oxide Fe/Si 5-8 198 25 20.7-12.9 – [15] Amorphous aluminum hydroxide 7 – RT 164 – [16] Fe/Si oxide 6.5 186 20.5 11.3 21.4 [17] Fe2O3 6 5.05 25 0.616 – [18] Al2O3 6 0.550 25 0.098 – [18] Iron oxide impregnated activated alumina 12 200 25 – 0.734 [19] Alum impregnated activated alumina 7 183 5.57 – [20] Activated alumina 7.6 370 25 – 0.180 [21] Mesoporous alumina 8 483 RT 61.3 – [22] Commercial activated alumina 8 322 RT 12.1 – [22] Meso-Al-400 7 396 RT 299 110 [23] Meso-80Al20Mg-400 7 422 RT 225 391 [23] Nanoparticle Fe-Al oxide 68.9 32 20.7 6.20 [24] Mg-Al hydrous oxide 7 290 22 220 – [25] Mesoporous alumina 4 312 52.8 39.1 – [26] Mesoporous alumina 6.6 312 25 36.6 – [26] Al2O3 (MA) 4 221 25 90.0 – This study Fe-Al2O3 (MA-5Fe) 4 104 25 62.0 – This study γ-Al2O3 (CA) 4 200 25 54.0 – This study Al2O3 (MA) 8 221 25 48.0 16.0 This study Fe-Al2O3 (MA-5Fe) 8 104 25 41.0 10.0 This study γ-Al2O3 (CA) 8 200 25 36.0 – This study 2 N. Inchaurrondo et al. Journal of Environmental Chemical Engineering 7 (2019) 102901 low cost, no chemical addition, regeneration capability and low energy carbonate (99.5%, Acros organics), sodium chloride (Panreac), po- consumption. The accessibility of the adsorption process allows its tassium nitrate (99%, Aldrich), sodium sulphate anhydrous (Sigma), implementation in individual homes and small community systems and sodium phosphate dibasic heptahydrate (Sigma), L(+) ascorbic acid makes it ideal to implement in developing countries. (Scharlau), sulphuric acid (95–97%, Scharlau), aluminum isopropoxide Several materials have been tested for arsenic removal such as ac- (≥ 98%, Aldrich), ferric nitrate nonahydrate Fe(NO3)3.9H2O tivated carbon, mineral oxides, zeolites, polymer resin or biological (AppliChem), nitric acid HNO3 (70% wt./wt., Cicarelli PA), hydro- materials [4,6], with the purpose of finding a regenerable adsorbent chloric acid HCl (37% wt./wt., Merck), tri-block copolymer Pluronic with great removal capacity and stability. Activated alumina (AA) has P123 (Aldrich, symmetric tri-block copolymer constituted of poly received substantial attention due to its relatively high surface area and (ethylene oxide) (PEO) and poly(propylene oxide) (PPO)), and anhy- high affinity towards several inorganic anions [7,8]. Results presented drous ethanol (99.5%, Cicarelli). by other authors, employing alumina or iron oxides, are collected and presented in Table 1, most of them only focused on As(V) removal. 2.2. Synthesis of mesoporous alumina adsorbents In spite its benefits, conventional AA has poorly defined pore structures and low adsorption capacities. An ideal adsorbent is expected Alumina powders were synthesized by using an adaptation and to have uniform accessible pores, a three dimensional pore system, high scaling-up over the evaporation-induced self-assembly methodologies surface area, fast adsorption kinetics and good physical and/or che- reported by Yuan and Morris separately [30]. During the synthesis, 20 g mical stability [7]. Therefore, the use of mesoporous alumina (MA) has of tri-block copolymer Pluronic P123 was dissolved in 200 mL of an- been proposed to improve the adsorption efficiency of AA due to its hydrous ethanol at room temperature and allowed to stir for 4 h (so- large surface area and narrow pore-size distribution, with uniform lution A). Meantime, solution B was prepared from 32 mL of nitric acid channels [23–26]. Moreover, they allow the incorporation of active and 100 mL of anhydrous ethanol. An amount of 0.2 mol of aluminum species through different methodologies, obtaining functional materials isopropoxide was added to solution B under vigorous stirring. Once the with greater reactivity and selectivity [27]. synthesis precursors were dissolved, the two solutions were combined Among different preparation methodologies, sol-gel is one of the in one-pot and 100 mL of anhydrous ethanol was used to thoroughly most preferred methods for the synthesis of mesoporous metal oxides transfer the aluminum isopropoxide solution. The final solution was since it provides a low-temperature route with excellent control over continuously stirred for 5 h at room temperature. The molar ratio mixing [28,29]. Based on the sol-gel methodology, a remarkable way of [Al3+]:[P123]:[EtOH]:[HNO3] in the final solution was fixed at achieving the growth of ordered MA structures is the solvent Eva- 1:0.017:30:2.5. In order to avoid the effect of humidity and to control poration Induced Self-Assembly (EISA). This one-pot strategy consists the hydrolysis rate of the aluminum alkoxide, the synthesis was carried in the self-assembly of a non-ionic triblock copolymer (P123), used as a out inside a dry glove box under N2 atmosphere. Afterwards, solvent structure directing agent, and the metal precursors, in an ethanolic evaporation of the homogeneous sol was performed in a drying oven at solution with nitric acid [30]. The incorporation of metallic species into 60 °C for 48 h (air atmosphere without stirring). The resulting xerogel the host structure during the MA synthesis may result in a strong in- was calcined in a muffle furnace with integrated air circulation by in- teraction between the doped metal and the alumina matrix, with the creasing temperature from room temperature to 400 °C (heating rate of development of large uniform mesopores [31]. This enhances the ac- 1 °C/min) and held at the final temperature for 4 h. The calcined sam- cessibility to surface sites and allows a homogeneous distribution of ples were ground in an agate mortar to reach the final mesoporous active sites, with improved thermal stability [31]. alumina powders, labeled as MA. The adsorption performance of the Although, there are several studies dealing with arsenic adsorption synthetized material was contrasted with the one of a commercial γ- by alumina materials, the improvements obtained from using MA, Al2O3 (SASOL, 200 m2/g), labeled as CA. synthetized by the EISA methodology, have not been deeply studied. In addition, the alumina matrix was doped with iron to form the Hence, in this contribution, we present a simple and reproducible mesoporous Fe-alumina adsorbent. This material was prepared by method to synthesize ordered mesoporous alumina, through an en- adding a Fe precursor, ferric nitrate, to solution A. The other steps of vironmentally friendly sol-gel route with P123 as soft template, to be the synthesis were the same as those given above. The total quantity of applied to the adsorption of As(V) and As(III) in contaminated waters. iron and aluminum was kept constant (0.2 mol) and the molar ratio Moreover, there has been great interest in the use of bimetals and bi- [Fe]:[Al] was adjusted to 0.064. The total iron content of the Fe-doped metal oxides for the removal of environmental contaminants with en- adsorbent resulted in 4.7% wt./wt. (labeled as MA-5Fe). couraging treatment efficiencies [10,19,32]. In particular, Fe based materials have shown higher affinity toward arsenate as compared to non-iron adsorbents [6,15,18,32]. Thus, the addition of Fe during the 2.3. Adsorbents characterization techniques MA synthesis was also studied. Subsequently, the performance of the synthesized materials was evaluated in the removal of As(V) and As(III) The characteristics of the alumina adsorbents were determined by species, focusing on the equilibrium properties, isotherms and kinetics different conventional techniques. adsorption behavior, at different pH values. A comparison with a commercial gamma alumina was also carried out to assess the im- 2.3.1. N2 physisorption provements obtained from the sol-gel synthesis. Additionally, the pre- The nitrogen adsorption and desorption isotherms at −196 °C were sence of interfering anions was also studied in order to evaluate a more measured using a Micrometrics ASAP-2020 instrument. Specific surface realistic water matrix. areas were calculated from the adsorption branches by the BET method, while pore size distribution were deduced from the desorption bran- 2. Methodology ches. Before analysis, each sample was degassed overnight at 120 °C under vacuum conditions. 2.1. Chemicals and reagents Analytical-grade reagents were used as supplied without further 2.3.2. X-ray diffraction (XRD) purification. The reagents used in this work were: sodium acetate tri- Powder XRD measurements were obtained with a PANalytical hydrate (> 99% ACS, Aldrich), acetic acid (> 99% ACS, Aldrich), so- X’Pert Pro diffractometer by using CuKα radiation (λ = 1.54056 Å). dium hydrogen arsenate heptahydrate (Aldrich), sodium metaarsenite The diffractograms were recorded over 10° < 2θ < 70° range and (98%, Aldrich), ammonium molybdate tetrahidrate (Fluka), sodium compared to the JCPDS files to confirm phase identities. 3 N. Inchaurrondo et al. Journal of Environmental Chemical Engineering 7 (2019) 102901 2.3.3. Transmission electron microscopy (TEM) 2.5. Adsorption experiments The microstructure of the alumina adsorbents was studied by TEM using a TEM JEOL 100 CX II instrument. Samples were prepared by Batch adsorption experiments were conducted to determine the As dispersing the powdered catalysts in ethanol and dropping the sus- adsorption equilibrium properties of the studied materials, by using pension onto a standard formvar-coated copper grid. known amounts of such adsorbents with different arsenic solutions (40 mL), adjusted at different pH values, under constant stirring. 2.3.4. Thermogravimetric analysis (TGA) All the adsorption experiments were carried out at room tempera- TGA of the synthetized xerogels was performed with a TGA Q500 V ture using polypropylene (PP) vessels on a rotary stirrer, during a 20.13 (TA instruments) thermobalance under air atmosphere at a maximum of 24 h (except for As(III) at initial pH of 11, where the ex- heating rate of 10 °C/min and a temperature interval between periment was extended up to 48 h). Stock solutions of 500 mg/L of As 25–700 °C. (V) and 100 mg/L of As(III) were prepared with Milli Q water and then diluted to various concentrations. The initial pH of arsenic solutions were adjusted as required with HNO3 (0.1 M or 1 M), NaOH (0.1 M) to 2.3.5. Point of zero charge (PZC) adjust at pH 11 or 0.02 M acetate buffer at pH 3.7. The PZC was determined by means of mass titration method. The pH of the resulting suspensions was monitored throughout the Portions of solid samples (0.05 g) were added to a stirred aqueous so- course of adsorption experiments; the final pH measured at the end of lution with initial pH 6. The pH of the system changes gradually and the experiments is the value reported as equilibrium pH (pHeq). approaches to a constant value, which equals to PZC. Measurements The adsorption capacity, QE (mg/L), of the adsorbent was calculated were performed at 25 °C in 25 mL of 0.005 mol/L NaCl aqueous solu- based on the following equation: tions, purged with 750 mL/min of N2. V QE = (C0 − CE )* m (1) 2.3.6. Temperature programmed desorption of pyridine (TPD) Acidic properties were determined by pyridine TPD as the probe where C0 (mg/L) and CE (mg/L) represent the initial and equilibrium molecule, using a Pyris 1 TGA instrument (Perkin Elmer). The ad- arsenic concentrations, respectively, V (L) is the volume of the arsenic sorbent samples were pretreated by heating from 50 to 400 °C (10 °C/ solution and m (g) is the mass of adsorbent. min, purged with air) and hold at 400 °C for 15 min to eliminate surface impurities. Then the samples were cooled to 120 °C and purged with N2, followed by the surface saturation with pyridine until constant weight. 2.5.1. Determination of adsorption isotherms The excess of probe molecule was then removed in N2 flow (120 min). The arsenate and arsenite adsorption isotherms were obtained at The pyridine covered samples were subjected to TG by heating up from different initial pH values (3.7, 7.5 and 11), by using adsorbent con- 120 to 400 °C (20 °C/min). centrations of 0.5 g/L (pH 3.7 and 7.5) and 5 g/L (pH 11). To evaluate the isotherm, different arsenic concentrations were 2.3.7. Fourier-transform infrared spectroscopy (FTIR) checked: 5, 10, 25, 50, 100, 250, 500 mg/L for As(V) and 5, 10, 25, 50, Different species in the sample were identified by FTIR/attenuated 100 mg/L for As(III). total reflection spectroscopy (FTIR/ATR Nicolet 6700). Spectra were recorded in the region of 4000–400 cm−1, at room temperature by 2.5.2. Determination of kinetic curves performing 64 scans with a resolution of 2 cm−1. The kinetic curves for As(V) and As(III) were also obtained at dif- ferent initial pH values (3.7, 7.5 and 11), by adding the same amounts 2.3.8. X-ray photoelectron spectroscopy (XPS) of adsorbent as mentioned in Section 2.5.1. The arsenic initial con- XPS analyses were performed using a multi-technique system centration was 100 mg/L for As(V) and 20 mg/L for As(III). Samples (SPECS) equipped with a dual Mg/Al X-ray source, and a hemispherical were collected at 0.5, 2, 5, 10, 24 h (and 48 h for As(III) at initial pH of PHOIBOS 150 analyzer operating in the fixed analyzer transmission 11). (FAT) mode. The spectra were obtained using an Al Kα radiation (h.ν = 1486.6 eV) operated at 200 W and 12 kV. The pass energy for the element scan was 30 eV. The working pressure in the analyzing 2.5.3. Influence of interfering anions and water matrix chamber was less than 8 × 10−10 kPa. The spectra of As 3d, O 1 s, C 1 s, The Cl−, SO42, NO3-, PO33- interference was checked using As(V) Fe 2p, Al 2p, Al 2 s and Auger As peak regions were recorded for each concentration of 20 mg/L, under pH 7.5 (near to real waters pH), with sample. The data treatment was performed with the Casa XPS program an adsorbent dose of 0.5 g/L and a molar ratio of [As]:[anion] = 1:25. (Casa Software Ltd, UK). The peak areas were determinate by integra- In order to test a more realistic water matrix, the adsorption test was tion employing a Shirley-type background. Peaks were considered a also carried out with an arsenate solution prepared with tap water, by mixture of Gaussian and Lorentzian functions. For the quantification of using the same adsorbent load and initial As(V) concentration. The tap the elements, we used the sensitivity factors provided by the manu- water sample was analyzed by Companyia d’Aigües de Sabadell S.A, facturer. revealing the presence of: Cl− (90.4 mg/L), Na− (37.4 mg/L), SO4-2 (68.7 mg/L), CaCO3 (174 mg/L). 2.4. Analytical measurements 2.5.4. Desorption of As(V) As(V) concentration was determined by the molybdene blue method Preliminary desorption experiments were performed by using sam- at 855 nm, using UV-1800 spectrophotometer [33]. As(III), Fe and Al in ples obtained from adsorption tests with 5 g/L of MA and MA-5Fe, in the supernatant were determined by Inductively Coupled Plasma Mass 50 mg/L of As(V) aqueous solutions (Milli Q water, initial pH 7.5), Spectrometry, ICP-MS (XSERIES 2 ICP-MS, Thermo Scientific, USA). stirred during 24 h. After the adsorption test, the samples were filtered Total Fe content of MA-5Fe solid sample was determined by a and dried at room temperature during 48 h, and then at 50 °C for 24 h. standard colorimetric test (FerroVer®Iron Reagent, HACH) over the To perform the As(V) desorption tests, mass samples of 0.05 g were digested solid samples in HNO3−HF. stirred during 5 h in 25 mL solutions of different initial pH values: pH was monitored using a pH meter micropH 2002 (Crison instru- HNO3 (0.01 M, pH 1.9), Milli Q water pH 8, NaOH (0.01 M, pH 11.5) ments, S.A., Spain), with a combined glass electrode. and NaOH (0.05 M, pH 12). 4 N. Inchaurrondo et al. Journal of Environmental Chemical Engineering 7 (2019) 102901 Fig. 2. XRD diffractograms of the tested adsorbents. Fig. 1. Thermogravimetric analysis of MA and MA-5Fe xerogels before calci- the Henry’s law at low pressure [35]. nation step. The mathematical expression for DR equation in the liquid phase system is: 2.6. Isotherm models and kinetics parameters 2 qe = QDR * e−KRD * ε (4) The adsorption results were fitted by using Langmuir, Freundlich, 1 Temkin, Dubinin–Radushkevich and Sip´s isotherm models. ε = R * T *ln(1 + ) Ce (5) 2.6.1. Langmuir isotherm where R and T represent the gas constant (8.314 J/mol K) and absolute The Langmuir equation (Eq. (2)) assumes: (1) fixed number of ac- temperature (K), respectively. The QRD (mg/g) is the maximum theo- cessible sites with the same energy, available on the adsorbent surface; retical adsorption capacity, KRD (mol2 J−2) is a constant related to the (2) reversible adsorption; (3) once an adsorbate occupies a site, no sorption energy and ε is the Polanyi potential. further adsorption can occur on that site; and (4) there is no interaction between adsorbate species [34]. 2.6.4. Temkin isotherm The Temkin isotherm equation (Eq. (6)) assumes that the heat of Qm * KL * Ce qe = adsorption of all the molecules in the layer decreases linearly with 1 + KL * Ce (2) coverage due to adsorbent–adsorbate interactions, and that the ad- Qm (mg/g) is the maximum saturated monolayer adsorption capacity of sorption is characterized by a uniform distribution of the binding en- an adsorbent and KL (L/mg) is a constant related to the affinity between ergies, up to some maximum binding energy [35]. The Temkin model is an adsorbent and adsorbate. Good adsorbents present high theoretical given by adsorption capacity Qm and a steep initial sorption isotherm slope (high R*T qe = *ln(A * Ce ) KL) [34]. B (6) Where R is the gas constant (8.314 J/mol K), T the temperature (K), B 2.6.2. Freundlich isotherm = (−ΔH) the variation of adsorption energy (kJ mol−1), and A is the The Freundlich isotherm (Eq. (3)) is the earliest known relationship Temkin equilibrium constant (L mg−1). describing the non-ideal and reversible adsorption, not restricted to the The Temkin equation is superior in the prediction of gas phase formation of a monolayer. This empirical model can be applied to equilibria, but adsorption in the liquid phase is a more complex phe- multilayer adsorption, with non-uniform distribution of adsorption heat nomenon. Numerous factors including pH, solubility of the adsorbate in and affinities over the heterogeneous surface [35]. the solvent and surface chemistry of the adsorbent, influence the ad- qe = KF * Ce n (3) sorption from liquid phase [37]. The adsorption coefficient, KF, characterizes the strength of ad- 2.6.5. Sip´s model sorption. The higher the value of KF is, the higher is the adsorbent Frequently, adsorption isotherms measured over broad adsorbate loading to be achieved. The exponent n is related to the energetic concentration ranges cannot be described exactly with only a single set heterogeneity of the adsorbent surface, determines the curvature of the of isotherm parameters or a single model adjustment. If the simple two- isotherm and speed of saturation. Freundlich isotherms with n < 1 parameter isotherms should be maintained, different parameter sets show relative high adsorbent loadings at low adsorbate concentrations. have to be applied for different concentration ranges; otherwise the Therefore, they are referred to as favorable isotherms, whereas iso- application of three-parameter isotherms is recommended [36]. The therms with n > 1 are characterized as unfavorable ones [36]. Sip´s model (Eq. (7)) is another empirical model that can simulate both Langmuir and Freundlich behaviors [36]. This equation describes the 2.6.3. Dubinin–Radushkevich (DR) isotherm saturation phenomenon at higher concentrations and at lower con- The DR isotherm (Eqs. (4) and (5)), is an empirical model initially centrations it reduces to the Freundlich isotherm [36]. conceived for the adsorption of subcritical vapors onto micropore solids Qa * (K a * Ce )u following a pore filling mechanism [35]. The model has successfully qe = 1 + (K a * Ce )u (7) fitted high solute activities and intermediate range of concentrations data, but has unsatisfactory asymptotic properties and does not predict Qa is the maximum theoretical adsorption capacity of the system (mg/ 5 N. Inchaurrondo et al. Journal of Environmental Chemical Engineering 7 (2019) 102901 Fig. 3. TEM images of the different tested alumina adsorbents. Table 2 Table 3 Summary of N2-physisorption outcomes. TPD of pyridine results. Sample PZC SBET Vpore dpore Sample Quantity of acid Density of acid Pyridine desorption (m2/g) (cm3/g) (nm) sites sites temperature (mmol/g) (mmol/m2) (ºC) MA 7.5 221 0.66 10.2 MA-5Fe 7.5 104 0.34 9.67 MA 0.130 0.0006 235 CA 7.5 200 0.50 7.15 MA-5Fe 0.200 0.0020 235 CA 0.097 0.0005 227 values at different pH values. Instead the affinity constant (Ka) can be varied to account for pH dependent sorption effects [38]. Since the heterogeneity index (u) is a material property, its value was assumed to be a constant at all pH values. 2.6.6. Assessment of kinetic parameters To evaluate the sorption rates, empirical pseudo-first order (Eq. (8)), pseudo-second order (Eq. (9)), Elovich (Eq. (10)) and the simpli- fied Intra-particle diffusion model (Eq. (11)) equations were adopted to model the experimental data. dqt = k1 *(qe − qt ) dt (8) dqt = k2 * (qe − qt )2 dt (9) 1 qt = *ln(1 + α * β * t ) β (10) qt = kp * t 0.5 + C (11) −1 −1 The parameters k1 (1/min) and k2 (g mg min ) are the pseudo- first order and the pseudo-second order rate constants, respectively; qt and qe are the arsenic adsorption amount (mg/g) at any time (t (min)) and at equilibrium, respectively. The Elovich coefficients, α and β, represent the initial adsorption rate (g mg−1 min−2) and the desorption coefficient (mg g−1 min−1), respectively. The Elovich equation can be derived from either a diffu- sion-controlled process or a reaction-controlled process [39] In the intra-particle diffusion model, kp (mg g−1 min−1) is the rate constant and C (mg/g) is a constant associated with the thickness of the boundary layer, where a higher value of C corresponds to a greater effect on the limiting boundary layer [40]. If a plot of qt against t0.5 is linear and passes through the origin, the adsorption is entirely governed by intra-particle diffusion. In contrast, if the intra-particle diffusion plot gives multiple linear regions, then the adsorption process is controlled by a multistep mechanism. Therefore, a nonlinear run of the q–t0.5 plot Fig. 4. N2 Physisorption results: isotherms at −196 °C (A) and pore size dis- over a broad time interval is not evidence for missing intra-particle tribution (B). diffusion impact on adsorption kinetics [36]. The isotherm and kinetics parameters were obtained with the g); Ka is the affinity constant for adsorption (L/mg) and u is the index of nonlinear method using the “Solver add-in” in Microsoft Office Excel. heterogeneity. Several error functions were calculated such as the hybrid fractional The Sip´s isotherm uses a single value to describe the maximum error function (HYBRID) (Eq. (12)), the average relative error (ARE) adsorption capacity and does not require fitted adsorption capacity (Eq. (13)) and the sum of the errors squared (ERRSQ), to predict the 6 N. Inchaurrondo et al. Journal of Environmental Chemical Engineering 7 (2019) 102901 Table 4 As(V) and As(III) maximum adsorption capacities for all materials tested. As(V): 0.5 g/L of adsorbent at pH0 3.7 and 7.5, and 5 g/L at pH0 11.5. As(III): 0.5 g/L of adsorbent at all initial pH of study. Time of experiment, 24 h; temperature, 25 °C. Maximum capacity QEm (mg/g) Maximum capacity Q’Em (mg/m2) Initial pH pHeq MA MA-5Fe CA MA MA-5Fe CA As(V) As(III) As(V) As(III) As(V) As(V) As(III) As(V) As(III) As(V) 3.7 4 90 – 62 – 54 0.41 – 0.60 – 0.27 7.5 8 48 16 41 10 36 0.22 0.07 0.39 0.10 0.18 11.5 10 15 7.2 10 7.6 – 0.07 0.03 0.10 0.07 – Fig. 5. As(V) adsorption at different pH equilibrium values (25 °C): A) MA isotherms, B) MA kinetic curves (C0 = 100 mg/L), C) MA-5Fe isotherms and D) MA-5Fe kinetic curves (C0 = 100 mg/L). Isotherms curves: solid line-Sips model and dash line-Freundlich model. The kinetic curves correspond to the Elovich model. optimum isotherm model [35]. 3. Results 100 (q − qe calc )2 3.1. Adsorbents characterization *∑ e isotherm m−p qe isotherm (12) After thermal treatment at 400 °C, TGA analysis confirmed that all q − qe calc organic residues from previous preparation steps were removed from 100 *∑ e isotherm the MA samples (see Fig. 1). TGA curves displayed three stages of m qe isotherm (13) weight loss, at different temperatures (Fig. 1). The total weight losses for the MA and MA-5Fe samples were c.a. 80%. The first step, the ∑ (qe isotherm − qe calc )2 (14) weight loss below 150 °C, is attributable to the removal of physically adsorbed species, such as water. The second step, located in the Where m is the number of experimental points, p the number of para- 150–225 °C range, corresponds to the decomposition of the P123 tem- meters to be determined, qe isotherm is the experimental value and qe calc is plate. Finally, the third step represents the remaining weight loss in the the value calculated using the model. 7 N. Inchaurrondo et al. Journal of Environmental Chemical Engineering 7 (2019) 102901 The CA physisorption isotherm displayed a different shape of pore structure, type IV with H2(b) hysteresis loop, which might indicate more complex pore structures in which network effects are important and can be related with pore blocking [41]. Surface reactivity is determined by the structural and coordinative arrangements around the surface metal centers. The point of zero charge (PZC) is a central concept in the adsorption of charged species and its position defines the affinity of the solid surface to the ionic species [42]. The PZC represents the pH at which the number of posi- tively charges species (M−OH2+) equals the number of negatively charged species (M–O−) nulling the net surface charge, where M stands for a surface metal center. The value of point of zero charge resulted in 7.5 for all samples, which is consistent with those reported in the lit- erature for similar metal oxide systems [42]. This result means that adsorbents surface might be positively charged and will adsorb/attract anions and/or negatively charged ligands from waters when working at pH < PZC. Metal oxides reactivity is determined by its acidity and basicity. Hydroxyl groups formed at the surface behave as Brönsted acid sites, whereas Lewis acids and Lewis bases are sites located on metallic ca- tions and coordinately unsaturated oxygens, respectively [8]. The acidity of mesoporous alumina samples was determined by TPD of pyridine (Table 3). The measurements revealed that Fe incorporation during the sol-gel synthesis of alumina (MA-5Fe) increased the quantity and density of acid sites in relation with MA. By contrast, the CA sample showed a lower concentration of surface acidic sites which might in- fluence the uptake of arsenic species. All samples desorbed pyridine at around 235 °C, which implies the same strength of acid sites. 3.2. Arsenic adsorption tests 3.2.1. pH effect As(V) and As(III) maximum adsorption capacities under the tested conditions are presented in Table 4. Initial pH values of 3.7, 7.5 and 11.5 led to final equilibrium values of 4, 8 and 10, respectively. The mesoporous alumina presents high alkalinity or internal buffer capa- city, which helps to maintain small pH changes [43]. As(V) isotherms at Fig. 6. As(III) adsorption at different pH equilibrium values (25 °C): A) MA and MA-5Fe isotherms and B) MA and MA-5Fe kinetic curves (C0 = 20 mg/L). equilibrium pH values of 4, 8 and 10 are presented in Fig. 5a–d for MA Isotherms curves correspond to the Freundlich model at pH 8 and the DR model and MA-5Fe samples. at pH 10. Kinetic curves at pH 8 correspond to the Elovich model. The three materials studied presented a PZC around 7.5 (see Table 2). The interaction between arsenic and the adsorbents depends not only on the surface charge but also in the speciation of the arsenic 225–500 °C range, which is assigned to the dehydroxylation of OHe molecules. The equilibrium dissociation constants of the two arsenic groups and conversion of hydrated alumina into transitional alumina oxidation states are quite different: H3AsO4(As (V)) pKa1 = 2.19, [31]. pKa2 = 6.94, pKa3 = 11.5, and H3AsO3(As(III)) pKa1 = 9.20, The powdered MA and MA-5Fe materials calcined at 400 °C were pKa2 = 12.1, pKa3 = 13.4. amorphous and no crystalline phases were detected by XRD (Fig. 2). During adsorption tests performed with initial pH 3.7 and 7.5, the TEM images showed the development of ordered mesoporous predominant As(V) solution species remained negatively charged structures with narrow channels of ca. 10 nm (Fig. 3). For comparison (H2AsO4− and or HAsO42-). In that pH range, the surface charge of the purposes, Fig. 3 also shows the micrograph of a commercial gamma- materials was positive to neutral. Therefore, the adsorption of As(V) alumina sample (CA), exposing significant differences on the alumina species occurred mostly via electrostatic interactions or ligand ex- microstructure which can influence the accessibility of the arsenic ad- change. Lower pH favors the ligand exchange reaction since H2O is sorbates. easier to displace from metal biding sites than OH- [44]. As it was In agreement with TEM results, N2 Physisorption measurements mentioned before, the pH values increased after As(V) adsorption, revealed the development of mesoporous materials with a BET surface under acidic and near-neutral media. This could be related to a proto- area of 221 m2/g for MA and 104 m2/g for MA-5Fe. Both samples dis- nation of the adsorbent surface and also to the ligand exchange reaction played type IV isotherms and H3 hysteresis loops, typical from meso- with either H2AsO4- or HAsO4-, accompanied by a release of OH-. porous materials [41] and a narrow pore size distribution centered on As the initial pH increased from 3.7 to 7.5, the adsorption capacity 10 nm (Table 2 and Fig. 4). However, it should be noted that the Fe- decreased due to the reduction of the electrostatic interaction. Also, as doping and/or the scaling-up of the EISA methodology impacted ne- pH increases, mono-charged H2AsO4− anions deprotonate to form gatively on the MA-5Fe porosity, showing a noticeable decrease on the double charged HAsO42− anions. If the ion exchange process is in- specific surface area and the pore volume. Table 2 and Fig. 4 also show volved, the double-charged anions may require more functional sites, the CA sample characterization results. The commercial alumina ex- resulting in less arsenic adsorption. hibited a BET surface area and pore volume of 200 m2/g and 0.5 cm3/g, Once pH > > PZC, the As(V) adsorption decreases much further, respectively, with an average pore width of 7.15 nm. These values were due to the Columbic inhibition between the negatively charged surface slightly lower than the obtained for the MA adsorbent. and the anionic species. In this pH range the As(V) adsorption may 8 N. Inchaurrondo et al. Journal of Environmental Chemical Engineering 7 (2019) 102901 Table 5 As(V) isotherms parameters (Langmuir. Freundlich. Temkin and DR). Model Parameters pHeq 4 pHeq 8 pHeq 10 MA MA-5Fe CA MA MA-5Fe CA MA MA-5Fe Langmuir Qm (mg/g) 86.07 59.25 50.23 46.15 46.1 40.95 14.53 9.73 Q'm (mg/m2) 0.39 0.57 0.25 0.21 0.44 0.2 0.066 0.094 KL (L/mg) 0.59 0.16 0.13 0.031 0.012 0.0092 0.04 0.037 R2 0.95 0.82 0.67 0.73 0.84 0.83 0.92 0.91 ERRSQ 690.62 1030.07 349.98 834.46 466.082 287.31 46.25 16.66 HYBRID 79.56 346.53 120.18 378.81 228.64 188.03 41.2 20.76 ARE 10.49 26.86 15.22 41.01 31.42 35.63 30.13 21.28 Freundlich KF (mgn−1.Ln/g) 33.76 23.8 19.46 9.58 4.57 2.66 2.37 1.69 KF (mgn−1.Ln/m2) 0.15 0.23 0.097 0.043 0.044 0.013 0.011 0.016 n 0.18 0.17 0.18 0.26 0.36 0.42 0.32 0.29 R2 0.93 0.94 0.94 0.95 0.94 0.94 0.95 0.92 ERRSQ 988.81 350.62 65.88 148.87 106.11 97.53 30.11 16.03 HYBRID 190.49 122.36 20.72 57.78 67.19 51.67 23.32 21.53 ARE 18.3 15.98 5.97 13.72 16.73 18.08 22.26 20.82 Temkin A (L/mg) 22.23 61.039 7.55 27.11 75 100 2.63 1.31 B (J/mol) 237.7 399.1 362.96 614.63 900 1000 1258.74 1659.66 R2 0.98 0.95 0.91 0.77 0.59 0.27 0.91 0.92 ERRSQ 315.11 285.84 99.52 1801.09 1157.37 1217.92 49.82 15.79 HYBRID 46.75 65.78 33.63 646.43 249.88 685.42 58.39 24.28 ARE 8.11 12.67 7.82 53.09 33.72 72.19 34.45 23.25 DR QmDR (mg/g) 87.2 54.63 44.94 40.88 35.46 29.26 12.39 8.56 Q'mDR (mg/m2) 0.39 0.53 0.22 0.18 0.34 0.15 0.056 0.082 KDR ((mol/J)2) 0.0000078 0.0000042 0.0000032 0.00016 0.00033 0.0004 0.000047 0.000085 R2 0.75 0.73 0.39 0.42 0.58 0.55 0.79 0.79 ERRSQ 3479.703 1526.06 638.88 1801.091 1217.4 755.92 116.97 41.25 HYBRID 811.5 460.7 198.31 696.43 505.67 440.38 115.28 70.32 ARE 33.21 31.8 17.82 53.09 50.14 57.1 46.85 37.7 Table 6 Table 7 As(V) Sip´s isotherm parameters. As(III) isotherms parameters (Langmuir. Freundlich. Temkin and DR). MA MA-5Fe CA Model Parameters pHeq 8 pHeq 10 Qa (mg/g) 111.7 98.7 71.2 MA MA-5Fe MA MA-5Fe Q'a (mg/m2) 0.51 0.95 0.36 Ka4 (L/mg) 0.1 0.0164 0.03549 Langmuir Qm (mg/g) 15 12 6.4 9.87 Ka8 (L/mg) 0.0013 0.00047 0.0011 Q'm (mg/m2) 0.068 0.12 0.029 0.095 Ka10 (L/mg) 0.00004 0.0000058 – KL (L/mg) 0.045 0.05 0.11 0.18 n 0.41 0.35 0.046 R2 0.84 0.91 0.88 0.95 R2 0.99 0.95 0.96 ERRSQ 34.93 11.85 9.28 5.15 ERRSQ 829.51 1653.23 312.07 HYBRID 66.38 23.13 33.79 5.25 HYBRID 69.61 81.52 61.86 ARE 22.25 18.82 35.02 7.72 ARE 16.23 18.015 16.78 Freundlich KF (mgn−1.Ln/g) 0.67 0.86 1.06 3.04 KF (mgn−1.Ln/m2) 0.003 0.0083 0.0048 0.029 n 0.74 0.61 0.45 0.27 R2 0.96 0.96 0.86 0.8 occur via H-bonding between the negative surface and HAsO42−, ERRSQ 8.28 4.53 10.78 22.53 competing with OH-. Also, even if a surface has a net negative charge, HYBRID 25.46 10.44 37.62 26.54 enough residual positive surface charges might still be present to attract ARE 14.34 11.33 36.41 19.58 some arsenic oxyanions. Temkin A (L/mg) 0.36 0.42 0.57 2.15 As(III) isotherms at the equilibrium pH 8 and 10 are presented in B (J/mol) 632.58 824.66 1292.32 1264.76 R2 0.84 0.89 0.94 0.89 Fig. 6a,b for MA and MA-5Fe samples. Under pH 9.2 (first pKa of ar- ERRSQ 34.82 12.85 4.24 12.5 senite), the non-ionic H3AsO3 is the dominant arsenite specie and HYBRID 111.12 44.32 10.45 11.65 therefore the interaction is only based on weak van der Waals forces. ARE 39.48 30.14 18.27 12.03 Hence, at initial pH 3.7, the arsenite adsorption resulted negligible and DR QmDR (mg/g) 14.33 10.93 6.23 8.27 Q'mDR (mg/m2) 0.065 0.11 0.028 0.08 in general adsorption uptakes of As(III) were much lower than the KDR ((mol/J)2) 0.000053 0.00004 0.0000067 0.0000021 obtained for As(V). Several studies reported an optimum arsenite re- R2 0.84 0.85 0.95 0.92 moval between pH 6–9 [2]; while higher pH values might cause a de- ERRSQ 36.19 18.45 3.55 8.49 cline in arsenite uptake due to electrostatic repulsion. Despite this, high HYBRID 124.03 86.65 18.63 12.22 arsenite removal was observed at pH 10 for MA-5Fe. A possible ex- ARE 51.08 41.59 24.85 13.26 planation might be the dissolution of the material under alkaline media and posterior co-precipitation of arsenic species with Fe oxy-hydroxides 3.2.2. Isotherms and adsorption kinetics (see Section 3.2.5) [6,45,46]. The As(V) isotherms parameters are presented in Table 5. According Taking into account that the affinity towards As(III) resulted lower to these results, the Temkin model only described properly the arsenate than the affinity shown for As(V), it is highly recommendable to pre- adsorption under acidic pH. The variation of the adsorption energy, B oxidize the contaminant (Fenton reaction, ozonation, etc.) to maximize = (−ΔH), resulted positive for all the studied materials, which in- the removal efficiency of the total amount of arsenic from contaminated dicates that the adsorption reaction is exothermic. On the other hand, water samples [6]. 9 N. Inchaurrondo et al. Journal of Environmental Chemical Engineering 7 (2019) 102901 Table 8 As(V) kinetics parameters. Model Parameters pHeq 4 pHeq 8 pHeq 10 MA MA-5Fe CA MA MA-5Fe CA MA MA-5Fe −1 Pseudo First Order k1 (min ) 0.0097 0.007 0.028 0.0042 0.0087 0.028 0.0019 0.0016 R2 0.88 0.89 0.96 0.82 0.72 0.78 0.94 0.98 ERRSQ 945.66 360.12 53.27 377.95 181.82 64.48 10.39 1.41 HYBRID 222.06 148.68 36.47 63.44 38.65 37.74 22.09 5.36 ARE 13.92 16.19 7.54 12.56 10.44 11.39 17.51 10.18 Pseudo Second Order k2 (g. mg−1. min−1) 0.00023 0.00017 0.0012 0.00017 0.00073 0.0024 0.00016 0.00015 R2 0.97 0.94 0.98 0.87 0.84 0.83 0.97 0.98 ERRSQ 251.32 199.68 16.28 270.4 106.79 47.63 5.95 1.59 HYBRID 54.58 81.49 11.19 50.39 22.17 23.33 11.21 4.24 ARE 6.56 12.21 3.97 12.59 7.07 9.1 11.77 8.15 Elovich α (g. mg−1. min−1) 3.94 1.43 85.44 0.34 0.68 5.29 0.051 0.025 β (mg/g) 0.077 0.094 0.24 0.15 0.32 0.52 0.32 0.36 R2 0.98 0.97 0.98 0.91 0.93 0.87 0.98 0.97 ERRSQ 137.38 105.23 22.77 185.24 43.25 37.16 3.42 2.14 HYBRID 29.96 37.18 12.38 61.62 10.44 13.89 4.97 3.75 ARE 5.61 7.68 3.88 18.17 5.67 7.86 7.24 6.07 Intraparticle Diffusion Ci (mg/g) 16.3 11.49 16.89 4.31 4.73 6.48 1.082 0.37 Ki (mg. g−1. min-0.5) 1.89 1.32 0.82 0.68 0.41 0.29 0.22 0.18 R2 0.83 0.88 0.57 0.91 0.89 0.66 0.95 0.94 ERRSQ 1373.03 383.86 506.88 189.38 73.73 97.95 8.33 4.91 HYBRID 158.64 66 126.19 167.82 14.42 13.25 5.11 11.49 ARE 11.81 10.13 14.28 30.11 6.06 6.14 5.36 12.44 Table 9 As(III) kinetics parameters. Model Parameters pHeq 8 pHeq 10 MA MA-5Fe MA MA-5Fe −1 Pseudo First Order k1 (min ) 0.008 0.0066 0.0009 0.0015 R2 0.75 0.84 0.98 0.94 ERRSQ 11.17 3.83 1.93 9.51 HYBRID 32.42 17 5.74 30.79 ARE 19.66 17.61 17.58 22.46 Pseudo Second k2 (g. mg−1. min−1) 0.002 0.0019 0.0002 0.00014 Order R2 0.84 0.91 0.91 0.95 ERRSQ 6.94 2.23 7.2 8.11 HYBRID 18.53 9.86 25.79 23.7 ARE 15.91 12.16 43.26 19.12 Elovich α (g. mg−1. min−1) 0.34 0.11 0.006 0.034 β (mg/g) 0.91 1.1 0.22 0.34 R2 0.92 0.95 0.99 0.96 ERRSQ 3.56 1.21 1.19 6.62 HYBRID 8.59 5.97 3.86 15.12 ARE 9.97 11.34 15.36 14.86 Fig. 7. Anions interference effect on As(V) adsorption capacity for MA and MA- Intraparticle Ci (mg/g) 1.7 0.91 0 0.62 5Fe ([As(V)]:[anion]mol = 1:25, pHeq = 8, T = 25 °C). Diffusion Ki (mg. g−1. min-0.5) 0.15 0.11 0.12 0.19 R2 0.84 0.89 0.94 0.95 ERRSQ 7.15 2.67 4.89 7.49 blockage while facilitating the access of metal ions to the adsorption HYBRID 8.77 7.93 73.61 10.74 sites [10,48]. ARE 8.68 13.71 98.42 12 Another interesting tool of analysis is the arsenic uptake related to the specific surface area of each adsorbent. The Freundlich constant per the fitted Freundlich parameters described more effectively the ar- square meter of surface area (K´F mgn−1*Ln/m2) was higher for the Fe- senate adsorption behavior over the whole pH range, as seen by other doped alumina, especially at acidic pH. While at neutral/alkaline pH, authors [3,7,14,23,43]. the K´F (mgn−1*Ln/m2) for MA and MA-5Fe resulted similar. According The Freundlich constant per gram of adsorbent (KF mgn−1*Ln/g) to these results, at pH < PZC arsenate species show increased affinity was higher in the case of MA sample, which might be connected to its towards Fe sites, as observed by other authors [6,15,18,32]. Taking into greater surface area in contrast to the MA-5Fe sample. However, MA account characterization results, the increased arsenate uptake might and CA samples have comparable surface areas, therefore the increased be also associated with a higher density of acid sites in contrast to MA adsorption capacity of the mesoporous alumina may be due to its par- and CA samples (Table 2). ticular structure and surface features, such as amorphous nature and For all the studied materials, the n values obtained were similar and higher concentration of Brönsted sites. As described previously in below the unity, which indicates favorable isotherms. Section 2.2, the MA sample was synthesized under conditions which Table 6 shows the calculated Sip´s parameters for As(V) adsorption minimized the formation of crystalline well-ordered surfaces, obtaining adjustment. In agreement with the other fitted isotherms, the maximum a high surface density of edge Al−OH groups [47]. Despite the similar Qa (mg/g) obtained was shown by the sample MA. The adsorption ca- pore diameters and surface areas in both adsorbents, the well-defined pacity (mg/g) of this sample was much higher than the value reached channels of P123-templated mesostructures, might prevent pore by the commercial alumina (CA) or the MA-5Fe sample. Again, the Q´a 10 N. Inchaurrondo et al. Journal of Environmental Chemical Engineering 7 (2019) 102901 The values of the adsorption affinity constant (Ka) increased at acidic pH due to an enhancement of the electrostatic attraction between the adsorbent surface and the arsenate species (higher adsorption affi- nity). The u values obtained for the alumina systems (MA, CA) were si- milar (around 0.44). While for the MA-5Fe sample resulted slightly lower (u = 0.35). As mentioned before for Freundlich isotherm, ex- ponent values below the unity provide information about the favor- ability of As(V) adsorption onto the materials tested. Table 7 shows As(III) isotherms parameters for the sol-gel synthe- tized materials. In this case, the Freundlich isotherm described properly the adsorption behavior at pH 8 and the DR model correlated better the arsenite removal, at pH 10. In general, the parameters obtained for both adsorbents (MA and MA-5Fe) resulted quite similar and slightly better for the MA-5Fe sample. In the case of As(III), Sip´s equation was not applicable, since there was no agreement between the experimental and calculated adsorption capacities. Fig. 8. FTIR of fresh and used samples of MA in the adsorption of As(V) under Kinetics studies were performed and the adsorption systems ap- different pH values (4, 8, 10). proach equilibrium after 5, 10 and 24 h, at equilibrium pH values of 4, 8 and 10, respectively. As was previously commented, at higher pH, the electrostatic attraction diminishes and the adsorption process slows down. Tables 8 and 9 present the kinetic parameters obtained. From these results, it can be established that As(V) and As(III) adsorption results fit better the Elovich model. 3.2.3. Ions interference effect on As(V) adsorption To emulate real waters conditions, the interference of different ions and different water matrices was assessed. The obtained results for arsenate removal are presented in Fig. 7. As was observed by other authors [14,16,17], the addition of NO3−, SO4-2 and Cl- did not cause a significant reduction of As(V) adsorption, by using MA. The complexes formed with these anions may be much weaker (outer sphere complex) than those formed between arsenate and MA [15,18]. In contrast with outer-sphere complexation, the adsorption through inner-sphere complexation is not greatly influenced by the alteration of the ionic strength of the solution [49]. However, from our results, the pH dependency of the adsorption process indicates a strong electro- Fig. 9. FTIR of fresh and used samples of MA-5Fe in the adsorption of As(V) static attraction mechanism. Therefore, it is possible that the arsenic under different pH values (4, 8, 10). oxyanions are predominantly adsorbed non-specifically by strong electrostatic attraction, followed by a strong inner-sphere complexation per square meter of surface area (mg/m2) resulted maximum for the between arsenic oxyanions and surface hydroxyl groups on the surface MA-5Fe sampleand MA sample displayed higher Q´a than the com- of the studied adsorbents [15,49]. mercial alumina. Consequently, as was discussed previously, the larger In the case of MA-5Fe, the effect of the interfering ions resulted adsorption capacity shown by the synthesized alumina material (MA) stronger, which may indicate that the bond between arsenate and MA- would be mostly related to its amorphous structure, higher concentra- 5Fe could be weaker than the bond formed with MA. The effect of tion of acid sites and the development of well-defined mesopores chloride has been ascribed to complexing between chloride and iron [10,49,47,48]. (III) as well as competition with arsenic species for adsorption sites Table 10 XPS results (binding energy (eV) and atomic ratio) for MA and MA-5Fe samples fresh and used in As(V) and As(III) adsorption tests. Atomic ratio Binding Energy (eV) As(V) As(III) Al Fe O As(V) As(III) Al Fe O MA – – 31.4 – 68.6 – – 73.4 – 530.0, 531.6 MA pHeq 4 500 mg/L As(V) 2.60 – 29.8 – 67.6 44.8 – 73.5 – 530.0, 531.6 MA pHeq 8 500 mg/L As(V) 0.700 – 30.5 – 68.8 44.5 – 73.6 – 530.5, 532.0 MA pHeq 10 500 mg/L As(V) 0.600 – 29.5 – 70.0 44.5 – 73.6 – 530.2, 531.6 MA pHeq 8 100 mg/L As(III) – 0.600 31.2 – 68.2 – 43.3 73.4 – 529.9, 531.5 MA-5Fe – – 30.3 1.00 68.7 – – 73.2 710.5 529.8, 531.3 MA-5Fe pHeq 4 500 mg/L As(V) 2.50 – 28.4 1.00 68.2 44.7 – 73.2 710.6 529.7, 531.3 MA-5Fe pHeq 8 500 mg/L As(V) 0.800 – 29.7 1.00 68.6 44.5 – 73.1 710.4 529.6, 531.2 MA-5Fe pHeq 10 500 mg/L As(V) 0.600 – 28.7 1.10 69.5 44.7 – 73.5 710.7 530.0, 531.6 MA-5Fe pHeq 8 100 mg/L As(III) – 0.500 30.1 1.10 68.2 – 43.6 73.6 710.8 530.1, 531.7 11 N. Inchaurrondo et al. Journal of Environmental Chemical Engineering 7 (2019) 102901 [50]. AlCl3. FTIR measurements performed in this study indicated a shift of When checking tap water, the As(V) adsorption increased for both the peak assigned to the stretching vibration Al-O (606 cm-1) to a lower materials. Studies reported that the addition of an electrolyte could value (559 cm-1) after As(V) complexation. In the MA-5Fe samples at increase the adsorption levels due to the depression of negative solid pH > PZC, attenuation of 670 cm-1 peak with intensification of surface charges in the alkaline region [51]. Chen et al [11], studied the 500–600 cm -1 peaks was observed, which may be related to the sub- effect of different ions during arsenic adsorption onto ferric im- stitution of −OH with As-O during coagulation/precipitation. pregnated volcanic ash. The authors observed that all of the in- A very small band appeared at 1030 cm−1 (Al-O-H vibration), vestigated coexisting cations (K+, Ca2+, and Fe3+) enhanced As(V) which slightly shifted (1040-1050 cm-1) after As(V) adsorption, espe- removal. Zhang et al. [39] observed the same behavior in the case of cially at pH > PZC, which could be attributed to the interaction be- Ca2+, and reported that metallic cations can link the adsorbent particle tween Al-O-H group and arsenic species [26]. with arsenate, forming a metal–arsenate complex or a me- In the case of MA, the broad band at 550 cm−1 was reduced after As tal–H2O–arsenate complex. In our study, the increased arsenate re- (V) adsorption, indicating a considerable reduction of metal oxygen moval in tap water may be related to the presence of co-existing cations bending vibration after the arsenic adsorption (arsenate may strongly such as Ca2+ and Na+, as observed in the study of Chen et al [11], interact with the surface oxide) [13]. when using tap and lake water. Studies reported that the presence of arsenate is resolved by a band In contrast with the insignificant interference observed in the pre- in the 856-866 cm−1 region [57]. However, it is difficult to distinguish sence of NO3−, SO4-2 and Cl−, phosphate had a very detrimental effect the As-O bond peak due to overlap with Al-O peak (840–850 cm−1) on As(V) adsorption, as it was reported by other studies [14,18,26,51]. [23,55]. Phosphate can compete with arsenate for binding sites on Fe2O3 and XPS measurements were also performed with samples before and Al2O3 due to the similarity of their structure (charge, same tetrahedral after As(V) adsorption (initial concentration 500 mg/L) at the equili- configuration). brium pH values of 4, 8 and 10 and after As(III) adsorption (initial concentration 100 mg/L) at the equilibrium pH 8. 3.2.4. FTIR & XPS analysis of adsorption mechanism XPS confirmed the presence of As in the adsorbent samples. New As According to several studies, the arsenic adsorption mechanism 3d peaks at 44.5–44.8 eV for As(V) and 43.3–43.6 eV for As(III) were involves electrostatic attraction and surface complexation processes. It observed after adsorption (Table 9). No change in the arsenic oxidation has been reported that As(V) forms predominantly inner-sphere bi- state was observed after the tests. dentate binuclear complexes with Al/Fe oxides [8,10,14,50–52]. For As In most experiments, the Al 2p and Fe 2p peaks shifted towards (III), the co-existence of inner and outer sphere adsorption complexes higher or lower binding energies after As adsorption tests (Table 10) was observed [8,50]. [23,54]. The binding energy change may be attributed to the interac- Inner-sphere surface complexes are created as a result of direct tion between arsenic and Al/Fe species present in the oxides due to chemical bond formation between the adsorbing anion (Lewis base) and ligand exchange reaction [56], and it is more noticeable at neutral to the surface (Lewis acid) [8]. These complexes are formed via a ligand basic pH since variations lower than ± 0.2 eV may account for ex- exchange reaction with a surface functional group, and as a result, no perimental error. Differences in the XPS peaks shift are connected to water molecules are present between surface functional groups and the different initial pH, arsenic speciation and surface complexes (inner- adsorbate ions. On the other hand, outer-sphere surface complexes are sphere: bidentate binuclear, bidentate mononuclear, monodentate bi- formed as a result of electrostatic attraction between opposite charges nuclear and monodentate mononuclear; or outer-sphere complexes) [36]. The model of outer-sphere complex formation presumes that ions [23]. Some authors reported that different kinds of surface complexes can also be bound to the surface sites by chemical bonds without losing can cause either a positive or negative Al 2p peak shift in materials such their hydration water. This means that a water molecule is located as Fe/Al hydroxides, when adsorbing different arsenate speciation, in- between the adsorbate ion and the adsorption site. Therefore, the dis- dicating that this peak displacement might not be necessarily consistent tance to the surface is larger and the binding strength is weaker in even though an As-O-Al bond is formed. comparison to inner-sphere complex formation [36]. For both adsorbents, the As/Al surface ratio decreased as pH in- In addition to sorption reactions, the formation of a small quantity creased, in concordance with their poorer adsorption performance at of newly formed secondary oxides and their co-precipitation with the high equilibrium pH values, due to the electrostatic repulsion between target metal(loid)s (e.g., FeAsO4.H2O, FeAsO4.2H2O, and Fe3(AsO4)2) is arsenic species and the adsorbents surface (Table 9). another important removal mechanism [53]. In order to investigate the interaction between Al/Fe oxides and As, FTIR measurements were performed with samples before and after As 3.2.5. Desorption studies (V) adsorption (initial concentration 100 mg/L) at the equilibrium pH Significant desorption levels were registered in the alkaline pH values of 4, 8 and 10 (Figs. 8 and 9). Both materials showed common range. At pH0 = 11.5, a desorption around 42% was obtained with MA peaks at 3400, 1635, 500 cm−1. The broad peak centered on and MA-5Fe materials. The higher concentration of hydroxyl anions led 3400 cm−1 is attributed to the presence of hydrogen bonded hydroxyl to the ion exchange desorption of arsenate ions. Nevertheless, a loss of groups which are also responsible for the absorption band at around Al2O3/Fe2O3 around 6% was detected after the desorption experiment. 1635 cm-1 ((HOH)bending mode for water). The broad peak centered At pH0 = 12, almost complete arsenate desorption was achieved. around 500-550 cm−1 is assigned to the stretch vibration of AleO However, around 16.5% of Al2O3 was dissolved in both adsorbents, (AlO6) [26]. with 12% loss of Fe2O3 in the MA-5Fe sample. In some of the spectra, the intensity of the 3400 cm−1 broad peak The arsenate desorption at high pH values may be also related to the decreased after As(V) adsorption, suggesting that hydroxyl groups are dissolution of the materials in the alkaline region. During the deso- replaced to bond As(V) [13,23]. This was observed in the case of MA, rption test, it was observed the formation of a heterogeneous emulsion, especially at lower pH, due to the stronger electrostatic interaction that could be easily eliminated through centrifugation or filtration. between H2AsO4− and the protonated hydroxyl groups. This also oc- Therefore, at basic pH values, the arsenic removal through dissolution curred in the case of MA-5Fe at pH 4. However, in sample MA-5Fe, at and co-precipitation with Al or Fe oxy-hydroxides cannot be discarded pH > PZC, the broad bands at 3400 and 500 cm-1 intensified. This may [6]. be related to species formed during iron/aluminum–arsenic co-pre- Negligible desorption was observed in the acidic/neutral pH range. cipitation caused by the formation of iron/aluminum (hydro)oxides Between pH0 3.5–7.5, the adsorbents were stable and no metal leaching [11]. Hu et al. [54] studied the coagulation of As(V) using FeCl3 and was detected. 12 N. Inchaurrondo et al. Journal of Environmental Chemical Engineering 7 (2019) 102901 4. Conclusions [9] S. Luther, N. Borgfeld, J. Kim, J.G. Parsons, Removal of arsenic from aqueous so- lution: a study of the effects of pH and interfering ions using iron oxide nanoma- terials, Microchem. J. 101 (2012) 30–36. Mesoporous alumina (MA) and mesoporous Fe doped alumina (MA- [10] M. Dai, L. Xia, S. Song, C. Peng, A. Lopez-Valdivieso, Adsorption of As(V) inside the 5Fe) were prepared by a facile sol-gel route at room temperature. The pores of porous hematite in water, J. Hazard. Mater. 307 (2016) 312–317. materials revealed an ordered pore structure with narrow channels of [11] R. Chen, Z. Zhang, Y. Yang, Z. Lei, N. Chen, X. Guo, C. Zhao, N. Sugiura, Use of ferric-impregnated volcanic ash for arsenate (V) adsorption from contaminated ca. 10 nm and amorphous walls. The synthetized adsorbents demon- water with various mineralization degrees, J. Colloid Interface Sci. 353 (2011) strated to be very effective for As(V) and As(III) removal, with max- 542–548. imum As(V) uptakes of 90 and 62 mg/g at an equilibrium pH of 4, for [12] Y. Mamindy-Pajany, C. Hurel, N. Marmier, M. Roméo, Arsenic (V) adsorption from aqueous solution onto goethite, hematite, magnetite and zero-valent iron: effects of MA and MA-5Fe, respectively. While at nearly neutral pH, maximum As pH, concentration and reversibility, Desalination 281 (2011) 93–99. (V) uptakes of 48 and 41 mg/L and As(III) uptakes of 16 and 10 mg/L, [13] T. Basu, K. Gupta, U.C. Ghosh, Performances of As(V) adsorption of calcined were obtained with MA and MA-5Fe, respectively. (250◦C) synthetic iron(III)–aluminum(III) mixed oxide in the presence of some groundwater occurring ions, Chem. Eng. J. 183 (2012) 303–314. The synthetized materials displayed an important arsenic uptake [14] V.K. Tchieda, E. D’Amato, A. Chiavola, M. Parisi, A. Chianese, M. Amamra, capacity in comparison to a commercial alumina sample and in relation A. Kanaev, Removal of arsenic by alumina: effects of material size, additives, and to similar metal oxides systems reported in the literature. This high water contaminants, Clean Soil Air Water 44 (5) (2016) 496–505. adsorption capacity is connected to its amorphous nature and enhanced [15] T. Mahmood, S.U. Din, A. Naeem, S. Mustafa, M. Waseem, M. Hamayun, Kinetics, equilibrium and thermodynamics studies of arsenate adsorption from aqueous so- surface and structural properties: uniform accessible mesopores and lutions onto iron hydroxide, Chem. Eng. J. 192 (2012) 90–98. surface acidity. [16] M.A. Anderson, J.F. Ferguson, J. Gavis, Arsenate adsorption on amorphous alu- The Fe doped adsorbent showed a higher adsorption capacity per minum hydroxide, J. Colloid Interface Sci. 54 (3) (1976) 391–399. [17] L. Zheng, Arsenic adsorption from aqueous solutions on an Fe(III)-Si binary oxide square meter of surface area (strong arsenate affinity). In this sense, it adsorbent, Water Qual. Res. J. Can. 39 (3) (2004) 267–275. would be desirable to optimize the sol-gel synthesis conditions to in- [18] Y. Jeong, F. Maohong, J. Van Leeuwen, J.F. Belczyk, Effect of competing solutes on crease its surface area and improve the adsorption capacity. arsenic(V) adsorption using iron and aluminum oxides, J. Environ. Sci. 19 (2007) 910–919. FTIR and XPS measurements evidenced the interaction between Al/ [19] S. Kuriakose, T.S. Singh, K.K. Pant, Adsorption of As(III) from aqueous solution onto Fe oxides and the arsenic species. The adsorption process may be based iron oxide impregnated activated alumina, Water Qual. Res. J. Canada 39 (3) on the ligand exchange adsorption mechanism and the electrostatic (2004) 258–266. [20] S.S. Tripathy, A.M. Raichur, Enhanced adsorption capacity of activated alumina by interaction related to the solution pH, arsenic speciation and samples impregnation with alum for removal of As(V) from water, Chem. Eng. J. 138 (2008) PZC. 179–186. Arsenate desorption was achieved under alkaline conditions. [21] T.S. Singh, K.K. Pant, Equilibrium, kinetics and thermodynamic studies for ad- sorption of As(III) on activated alumina, Sep. Purif. Technol. 36 (2004) 139–147. However, the stability of the materials may be compromised depending [22] M.-J. Yu, X. Li, W.-S. Ahn, Adsorptive removal of arsenate and orthophosphate on the alkaline solution concentration. anions by mesoporous alumina, Microporous Mesoporous Mater. 113 (2008) In general, arsenic isotherms adjusted well to the Freundlich model 197–203. and more accurately to the three parameters Sip´s model. The kinetics [23] W. Li, D. Chen, F. Xia, J.Z.Y. Tan, P.-P. Huang, W.-G. Song, N.M. Nursam, R.A. Caruso, Extremely high arsenic removal capacity for mesoporous aluminium results fitted the Elovich model. magnesium oxide composites, Environ. Sci. Nano 3 (2016) 94. Under the studied operating conditions, the addition of interfering [24] J. Maiti, S. De Kumarbasu, Fe-Al nano-oxide prepared by sol-gel method using ions (NO3−, SO4-2 and Cl−) did not cause a significant reduction of As precursor of HCl digested liquid fraction of laterite: arsenic adsorption perfor- mance, Int. J. Nanosci. 10 (5) (2011) 1173–1177 4. (V) adsorption using MA. However, phosphate had a very detrimental [25] N. Chubar, New inorganic (an)ion exchangers based on Mg–Al hydrous oxides: effect on As(V) adsorption due to its similar structure with arsenate. (Alkoxide-free) sol–gel synthesis and characterisation, J. Colloid Interface Sci. 357 Additionally, the use of a tap water matrix increased arsenate adsorp- (2011) 198–209. [26] C. Han, H. Pu, H. Li, L. Deng, S. Huang, S. He, Y. Luo, The optimization of As(V) tion in both materials, indicating a favorable perspective for the use of removal over mesoporous alumina by using response surface methodology and these materials in the treatment of real effluents. adsorption mechanism, J. Hazard. Mater. 254–255 (2013) 301–309. [27] C. Márquez-Alvarez, N. Žilková, J. Pérez-Pariente, J. Čejka, Synthesis, character- ization and catalytic applications of organized mesoporous aluminas, J. Catal. Rev. Acknowledgements 50 (2008) 222–286. [28] M. Trueba, S.P. Trasatti, γ-Alumina as a support for catalysts: a review of funda- This work was supported by CONICET (PIP-0575), UNMdP, mental aspects, Eur. J. Inorg. Chem. 17 (2005) 3393–3403. [29] S.J. David, S.J.H. Justin, Metal Oxide Catalysis, WILEY-VCH Verlag GmbH & Co. ANPCyT (PICT1992-2014) (Argentina), the Spanish Ministry funding KGaA, 2009. (CTM2015-65414-C2-1-R), Marie Skłodowska-Curie Actions: Research [30] Q. Yuan, A. Yin, C. Luo, L. Sun, Y. Zhang, W. Duan, H. Liu, C. Yan, Facile synthesis and Innovation Staff Exchange (RISE): “Advanced multifunctional ma- for ordered mesoporous γ-aluminas with high thermal stability, J. Am. Chem. Soc. terials applied to remove arsenic in Argentinian groundwater” 130 (2008) 3465–34729. [31] W. Cai, J. Yu, C. Anand, A. Vinu, M. Jaroniec, Facile synthesis of ordered meso- (NANOREMOVAS). porous alumina and alumina-supported metal oxides with tailored adsorption and framework properties, Chem. Mater. 23 (2011) 1147–1157. References [32] F. Fu, Z. Cheng, J. Lu, Synthesis and use of bimetals and bimetal oxides in con- taminants removal from water: a review, RSC Adv. 5 (2015) 85395–85409. [33] R.K. Dhar, Y. Zheng, J. Rubenstone, A. van Geen, A rapid colorimetric method for [1] Arsenic in Drinking-water Background Document for Development of WHO measuring arsenic concentrations in groundwater, Anal. Chim. Acta 526 (2004) Guidelines for Drinking-water Quality, © World Health Organization, 2011. 203–209. [2] M. Jang, S.-H. Ming, J.K. Park, E.J. Tlachac, Hydrous ferric oxide incorporated [34] Langmuir, The adsorption of gases on plane surfaces of glass, mica and platinum, J. diatomite for remediation of arsenic contaminated groundwater, Environ. Sci. Am. Chem. Soc. 40 (1918) 1361–1403. Technol. 41 (2007) 3322–3328. [35] K.Y. Foo, B.H. Hameed, Review: insights into the modeling of adsorption isotherm [3] Y.-F. Pan, C.T. Chiou, T.-F. Lin, Adsorption of arsenic(V) by iron-oxide-coated systems, Chem. Eng. J. 156 (2010) 2–10. diatomite (IOCD), Environ. Sci. Pollut. Res. 17 (2010) 1401–1410. [36] Worch, Adsorption Technology in Water Treatment Fundamentals, Processes, and [4] M.I. Litter, M.E. Morgada, J. Bundschuh, Review: possible treatments for arsenic Modeling, 1st ed., de Gruyter, Göttingen, 2012. removal in Latin American waters for human consumption, Environ. Pollut. 158 [37] J. Febrianto, A.N. Kosasih, J. Sunarso, Y.-H. Ju, N. Indraswati, S. Ismadji, Review: (2010) 1105–1118. Equilibrium and kinetic studies in adsorption of heavy metals using biosorbent: a [5] H. Malik, Z.M. Khan, Q. Mahmood, S. Nasreen, Z.A. Bhatti, Review: perspectives of summary of recent studies, J. Hazard. Mat. 162 (2–3) (2009) 616–645. low cost arsenic remediation of drinking water in Pakistan and other countries, J. [38] G.P. Jeppu, T.P. Clement, A modified Langmuir-Freundlich isotherm model for si- Hazard. Mat. 168 (2009) 1–12. mulating pH-dependent adsorption effects, J. Contam. Hydrol. 129–130 (2012) [6] K. Henke, Arsenic: Environmental Chemistry, Health Threats and Waste Treatment, 46–53. Wiley, Chippenham, 2009. [39] J. Zhang, R. Stanforth, Slow adsorption reaction between arsenic species and goe- [7] D. Mohan, C.U. Pittman Jr., Arsenic removal from water/wastewater using ad- thite (r-FeOOH): Diffusion or heterogeneous surface reaction control, Langmuir 21 sorbents—a critical review, J. Hazard. Mater. 142 (2007) 1–53. (2005) 2895–2901. [8] B. Kasprzyk-Hordern, Chemistry of alumina, reactions in aqueous solution and its [40] W.J. Weber Jr., J.C. Morris, J. Sanit, Kinetics of adsorption on carbon from solution, application in water treatment, Adv. Colloid Interface Sci. 110 (2004) 19–48. J. Sanitary Eng. Div. Am. Soc. Civil Eng. 89 (1963) 31–38. 13 N. Inchaurrondo et al. Journal of Environmental Chemical Engineering 7 (2019) 102901 [41] M. Thommes, K. Kaneko, A.V. Neimark, J.P. Olivier, F. Rodriguez-Reinoso, bentonite with manganese oxides and poly(dimethyldiallylammonium chloride), J. J. Rouquerol, K.S.W. Sing, Physisorption of gases, with special reference to the Environ. Chem. Eng. 6 (1) (2018) 156–168. evaluation of surface area and pore size distribution, IUPAC Tech. Rep. Pure Appl. [50] D.E. Giles, M. Mohapatra, T.B. Issa, S. Anand, P. Singh, Review: iron and aluminium Chem. 87 (9-10) (2015) 1051–1059. based adsorption strategies for removing arsenic from water, J. Environ. Manage. [42] M. Kosmulski, pH-dependent surface charging and points of zero charge III. Update, 92 (2011) 3011–3022. J. Colloid Interface Sci. 253 (2002) 77–87. [51] W. Zhang, P. Singh, E. Paling, S. Delides, Arsenic removal from contaminated water [43] L.M. Camacho, S. Ponnusamy, I. Campos, T.A. Davis, S. Deng, S.I.S. Flora (Ed.), by natural iron ores, Miner. Eng. 17 (2004) 517–524. Handbook of Arsenic Toxicology: Evaluation of Novel Modified Activated Alumina [52] K. Mandal, K.T. Suzuki, Arsenic round the world: a review, Talanta 58 (2002) as Adsorbent for Arsenic Removal, Elsevier, 2015, pp. 121–136. 201–235. [44] E. García-Sanchez, F. Alvarez-Ayuso, Rodriguez-Martin, Sorption of As(V) by some [53] M. Komárek, A. Vanëk, V. Ettler, Review: chemical stabilization of metals and ar- oxyhydroxides and clay minerals. Application to its immobilization in two polluted senic in contaminated soils using oxides-a review, Environ. Pollut. 172 (2013) 9–22. mining soils, Clay Miner. 37 (2002) 187–194. [54] C. Hu, Q. Chen, H. Liu, J. Qu, Coagulation of methylated arsenic from drinking [45] X. Carrier, E. Marceau, J.F. Lambert, M. Che, Transformations of γ -alumina in water: influence of methyl substitution, J. Hazard. Mat. 293 (2015) 97–104. aqueous suspensions 1. Alumina chemical weathering studied as a function of pH, J. [55] S. Goldberg, C.T. Johnston, Mechanisms of arsenic adsorption on amorphous oxides Colloid Interface Sci. 308 (2007) 429–437. evaluated using macroscopic measurements, vibrational spectroscopy, and surface [46] G. Lefevre, M. Duc, P. Lepeut, R. Caplain, M. Fedoroff, Hydration of γ -alumina in complexation modeling, J. Colloid Interface Sci. 234 (2001) 204–216. water and its effects on surface reactivity, Langmuir 18 (2002) 7530–7537. [56] H. Li, L. Zhang, H. Dai, H. He, Facile synthesis and unique physicochemical prop- [47] M.B. McBride, L.G. Wessellnkt, Chemisorption of catechol on gibbsite, boehmite, erties of three-dimensionally ordered macroporous magnesium oxide, gamma-alu- and noncrystalline alumina surfaces, Environ. Sci. Technol. 22 (1998) 703–708. mina, and ceria-zirconia solid solutions with crystalline mesoporous walls, Inorg. [48] L. Mercier, T.J. Pinnavaia, Access in mesoporous materials: advantages of a uniform Chem. 48 (2009) 4421–4434. pore structure in the design of a heavy metal ion adsorbent for environmental re- [57] L. Sigg, W. Stumm, Aquatische Chemie [Aquatic Chemistry], vdf Hochschulverlag, mediation, Adv. Mater. 9 (6) (1997) 500–503. Zurich, 2011. [49] J. Hua, Adsorption of low-concentration arsenic from water by co-modified 14