Synthesis and characterization of aluminum-based adsorbent and application in fluoride removal from aqueous solution

A novel adsorbent was obtained by a facile precipitation method and was used for fluoride removal from aqueous solution. Mineralogical and physicochemical characterization of the adsorbent was carried out by X-Ray Diffraction (XRD), X-Ray Fluorescence (XRF), Energy Dispersive X-Ray attached to Scanning Electron Microscopy (SEM-EDX), BET Specific Surface Area(SSAN2BET) analysis and Fourier-Transform Infrared Spectrometry (FTIR). The effect of various operational parameters such as contact time, initial fluoride concentration, (20-160 mg L) adsorbent dose (1-6 g L) and initial pH solution (3-11) was evaluated in batch procedures at room temperature (25±2°C). The results of the batch adsorption experiments proved that 24 h of contact time was sufficient for attaining equilibrium. The maximum wastewater defluoridation (84.91%) was obtained for 40 mg L and 3 g L of initial fluoride concentration and adsorbent dose, respectively. It appears that there was no significant effect on the F removal over a wide range of pH 3-11. Kinetic studies revealed that fluoride adsorption fitted well to pseudo-second-order. The adsorption isotherm of fluoride sorption indicated that the maximum adsorption capacity was noted to be 43.29 mg g. Batch adsorption data was better described by Langmuir isotherm confirming monolayer adsorption with homogenous distribution of active sites and without interaction between adsorbed molecules. The obtained results indicated that the ion exchange is probably the main mechanism involved in the F adsorption by the aluminium-based adsorbent.


Introduction
Fluoride, a naturally occurring element present in the soil, water, food and several minerals composition 1,2 . It is an essential trace element for the human body for the dental calcification enamel and maintenance of healthy bones. Furthermore, fluoride is widely used in the organic drug industry 3,4 , fluorinated polymers 5,6 , surfactants 7,8 , battery components 9 , etc. However, fluoride contamination has been recognized as a serious problem worldwide, an excess of this element can lead to ecological damage 10 . This excess is mainly due to the dissolution of natural minerals like fluorapatite and fluorite in the aquatic environment 11 . Volcanoes can also lead to the release of gases containing hydrogen fluoride into the atmosphere before its transfer to the aquatic ecosystem 12,13 . In addition to natural sources, fluoride may also be found in industrial wastes such as semiconductor industries [14][15][16] the metal surface treatment 17 , the fertilizers production 18 , the application of the fertilizers and pesticides in the agriculture, and glass manufacturing 19,20 . The discharge of these industrial effluents can lead to an increase in the fluoride level in the receiving aqueous media (ex: groundwater table) where they are transferred by infiltration which can lead to environmental and health impacts [21][22][23][24] . In fact, recent literature reports that an excessive fluoride amount can produce toxic effects in animals and plants such a way that the fluoride ions act as enzyme poisons, inhibiting enzymatic activity and interrupting metabolic processes like the glycolysis and plant protein synthesis 25 .
Moreover, the high concentrations of fluoride can inhibit or increase the algae population growth and lead to a decrease in oxygen levels in aquatic systems 22 . Recently many scientific reports seem to suggest that fluoride intake may be associated with a number of human health problems [26][27][28] , for instance osteoporosis 29,30 , arthritis, bone fragility 31 , brain damage 32 , etc. Due to these impacts on human health and environment, a number of international regulators have issued regulations to control the fluoride levels in water 33,34 . Different processes have been studied to remove excessive fluoride from water namely, biological processes (plants or microorganisms) 35,36 and physicochemical processes  such  as  fluidized  bed  precipitation  37 ,  electrocoagulation-flotation 38,39 , ion-exchange 40,41 ,  electrochemical methods 42 and membrane processes  (reverse osmosis 43,44 ,   electrodialysis 45,46 , Donnan dialysis 47 , nanofiltration 48 , etc.). However, each of these methods presents its drawbacks, such as a long disposal time, the use of dangerous and expensive reagents 49 . Among these methods, adsorption is mentioned as the most efficient and widely used fundamental approach in water defluoridation processes 50 . It is characterized by its simplicity and availability of a wide variety of adsorbents. The adsorption capacity and adsorbents affinity are determined by the textural and physical properties as well as their surface chemical nature 51 . A large variety of adsorbents has been reported in literature, such as iron-based materials 52 , kaolinite 53 , bentonite 54 , lignite 55 , calcite 56 , tricalcium phosphate 57 , hydroxyapatite 58 , red mud 59 , montmorillonite 60 , Alumina has been reported in the literature as an active support for water defluoridation due to its sorption properties, ion exchange potential and low cost 61,62 . However, its adsorption capacity is not always satisfactory, mainly due to the small specific surface area. For this reason, the modification of its surface, by molecules having higher fluoride adsorption, has recently been investigated 63 . Our research aims at the removal of fluoride ions from aqueous solutions through adsorption by the aluminum-based adsorbent. The investigations were focused on the dynamic adsorption behavior of fluoride on the surface of the adsorbent, including isotherms studies and kinetics, in batch experiments. The significant novelty of this study is the simple preparation of a novel aluminum-based adsorbent that increases its affinity for fluorine and its stability in different pH.

Preparation of adsorbent
Aluminum-based particles are synthesized by precipitation method 64 . Aluminum nitrate and sodium hydroxide as the basic raw material were used for the process. The uniform aluminum solution was prepared by dissolving 375,13 g aluminum nitrate (Al(NO3)3,9H2O) into deionized water. Then, 800 mL of a 5 M NaOH solution was added. The aluminum nitrate reacted with sodium hydroxide, at room temperature, in a ratio of 1:4. The nucleophilic substitution of the OH group makes it possible to prepare Al(OH)4 with sodium nitrate and water. The obtained precipitate was filtered and dried in a draught drying cabinet at 105°C for 12 h. The obtained powder was ground, sieved through 1 mm sieve, and then calcined in a muffle furnace at 750°C for 4 h in the presence of air at a heating rate of 10°C.min -1 . The theoretical equations for the adsorbent synthesis can be written as Eqs. 1 and 2 64 . Al(NO3)3,9H2O + 4NaOH → Na(Al(OH)4) + 3NaNO3 + 9H2O (1) Na(Al(OH)4) → Na(AlO2) + 2H2O (2)

Characterization of adsorbent
An accurate structural and morphological characterization of adsorbent was determined by X-Ray Diffraction (XRD X'Pert PRO Panalatycal) and X-Ray Fluorescence (XRF, Oxfordmdx 1000). The Specific Surface Area was determined by the BET method using nitrogen adsorption (SSAN2BET) (Flowsorb II 2300, Micromeritics). Energy Dispersive X-ray attached to Scanning Electron Microscopy (SEM-EDX) was used to observe the morphological features of synthetic adsorbent and to determine the spot element analysis of these latter. The functional groups present in adsorbent were determined by FT-IR using a Fourier Transform Infrared spectrophotometer, (Model Bruker Vertex 70. FTIR) spectra were recorded between 4000 cm -1 and 400 cm -1 .

Batch adsorption experiments
The batch experiments consisted analysing the effect of controlling parameters such as contact time, initial Fconcentration, adsorbent dose and initial pH solution (the pH adjustment was carried out by using 0.1 N (HCl) or 0.1 N (NaOH)) on Fadsorption. All the experiments were conducted at room temperature of 25±2°C in 500 mL plexiglass reactor. A specific amount of adsorbent was added to the fluoride-doped synthetic solution. The reaction mixture was continuously blended by an electric rod stirrer at 350rpm. At equilibrium, the liquid phase is recovered in an Erlenmeyer flask after vacuum filtration through Whatman No. 42 filter paper for Fanalysis, The equilibrium study is performed by varying initial Fconcentrations from 20 to 160 mg L -1 , while maintaining the constant adsorbent dosage and the contact time at 3 g L -1 and 24 h, respectively, for each sample. The effect of adsorbent dosage is studied by varying the mass of adsorbent from 1 to 6 g L -1 in a sample of 40mg L -1 initial Fconcentration at natural pH (7±0.2). The effect of pH is also studied by conducting experiments for a variation of pH from 3 to 11 by keeping the initial Fconcentration and the adsorbent dose at 40 mg L -1 and 3g L -1 , respectively.

Fluoride removal calculation
The specific amount of the adsorbed F -, Qe (mg g -1 ), was calculated according to Eq. (3): Where, Qe is the adsorption capacity (mg g -1 ) in the solid at equilibrium; Co, Ce are initial and equilibrium concentrations of F -(mg L -1 ), respectively; V is the volume of the aqueous solution (L) and W is the mass (g) of adsorbent used in the experiments.
The adsorption removal efficiency (ARE), at equilibrium, was calculated by using Eq. (4):

Modeling studies 2.7.1. Kinetic models
The pseudo-first-order is a kinetic model described by the following Lagergren Eq. (5) 67 : The linear form of pseudo-first order kinetic model can be expressed by Eq. (6): Where, e and t are the amount of Fadsorbed (mg g -1 ) at equilibrium and at a time 't', respectively. K1 (min -1 ) represents the rate constant of pseudofirst-order adsorption reaction. A straight line of log( e − t ) against t suggests the applicability of these kinetic models. Both e and K1 can be determined from the intercept and slope of the curve, respectively.
The linear form of the pseudo-second-order kinetic model can be expressed by Eq. (7) 68 : where K2 is the rate constant for pseudo-secondorder reaction (g mg -1 min -1 ). e and t are the amounts of Fadsorbed at equilibrium and at any time 't' (mg g -1 ), respectively. The straight-line plot of t/ t against t for the kinetic data gives the values for e and K2 from the slope and intercept, respectively.

2.7.2.. Adsorption isotherms
The isotherm equations Langmuir and Freundlich isotherm models have been used in the present study, to describe the equilibrium data. The Langmuir equation can be described in the following Eq. (8) 69 : The linearized form of Eq. (6) can be written as Eq. (9): where Ce is the equilibrium concentration of fluoride ions (mg L -1 ), Qe is a solid phase concentration of fluoride ions (mg g -1 ), Qm (mg g -1 ), and KL (L mg -1 ) are empirical constants, can be evaluated from the slope and intercept of the linear plot of 1/Qe against 1/Ce.
The Freundlich is expressed mathematically in linear form as it is represented in Eq. (10) 70 : Where, KF (mg g -1 ) and 1/n are Freundlich constants related to adsorption capacity and adsorption intensity, respectively. Those constants are obtained from the intercept and slope of logQe versus log Ce linear plot, respectively.

Characterization of adsorbent 3.1.1. X-Ray Diffraction (XRD) analysis
The structural characterization of synthetic adsorbent was carried out by X-Ray Diffraction (XRD) as shown in Figure 1. XRD pattern of adsorbent showed three major well-crystallized phases (sodium nitrate (NaNO3), sodium aluminum oxide (Na (AlO2)) and sodium carbonate (Na2CO3)) with well-resolved peaks.

X-Ray Fluorescence (XRF) analysis
The chemical constituents of adsorbent are shown in Table 1. The analysis revealed that alumina (Al2O3) was the main component at 44.97%, followed by sodium oxide (Na2O) at 9.45%, confirming that synthetic adsorbent is based on sodium aluminum oxide.

Energy Dispersive X-ray attached to Scanning Electron Microscopy (SEM-EDX) analysis
The results of SEM-EDX analysis of synthetic adsorbent are presented in Figures 2, 3. The SEM pictures showed surface morphology and grain distribution. The EDX spectrum implied that the composite is made up of uniformly distributed carbon, oxygen, sodium and aluminum all over the surface as shown in Figure 3.

BET Specific Surface Area (SSAN2BET) analysis
The BET results show that the aluminum basedadsorbent presents an important surface area (≃10.0059 m 2 g -1 ) comparing to most of the activated alumina adsorbents with some exceptions.

Fourier-Transform Infrared Spectrometry (FTIR) analysis
The FTIR spectra of the aluminous support, before and after adsorption (Figure 3), showed common bands corresponding to the basic structure (AlO2) such as two bands at 1371 cm −1 and 1236 cm −1 which most probably correspond to the symmetrical and antisymmetric elongation of Al=O group.
Furthermore, two bands at 966 cm −1 and 511 cm -1 correspondings to the deformations of the group Al=O 71 . While the appearance of two bands at 1742 cm −1 and 3428 cm -1 are characteristic of the OH group of the H2O molecule. The comparison of the FTIR spectra did not show any difference between the infrared spectra of the adsorbent (Na (AlO2)) before and after the adsorption tests. Nevertheless, if we look carefully (the spectrum in blue), we can notice that there is an appearance of small bands at 507.58cm −1 and 554.94cm −1 correspondings to the Al-F bond 72 , which confirms the adsorption of Fions by our adsorbent through a chemisorption mechanism.

The effect of contact time
The effect of contact time on the percentage of Fadsorbed was investigated to ensure equilibrium state between aqueous Fand the adsorbent.  Figure 4 shows the progression of adsorption reaction, the percentage of the Fadsorbed on to the adsorbent after different contact times. It was observed that with a fixed amount of the adsorbent, the percentage of Fadsorbed increased with contact time and began to reach equilibrium after 24 h of reaction. Consequently, the contact time fixed for further adsorption experiments was 24 h.

The effect of initial fluoride concentration
The effect of initial concentration on the percentage of Fremoval was carried out to determine the maximum adsorption capacity of the synthetic adsorbent. The adsorbent dose and initial pH solution were fixed at 3 g L -1 and 7±0.2, respectively. As shown in Figure 5, the adsorption capacity reached stability at a high initial concentration (100 mg L -1 ). This can be explained by saturation of the available active adsorption sites. Though, the percentage of Fremoval decreased with an increase in the initial concentration. This decrease is due to the presence of more Fion in solution at higher initial Fconcentration. However, the efficient Fremoval at a low initial concentration was because of the important ratio of surface active sites to total Fions present in solution.  Figure 6 shows that the percentage of Fremoval increased slightly with increasing adsorbent dose up to 3 g L -1 . The increase of percent Fremoval with the increase of the adsorbent dose is due to the availability of sufficient adsorption sites. However, the addition of the adsorbent dose did not show any considerable increase in the percentage of Fremoval. This may be probably due to the overlapping phenomenon of the active sites at a higher adsorbent dose resulting in a reduction of surface area 73 .  Figure 7 illustrates the evolution of the percentage of Fremoval over the pH range of 3-11. It was observed that the percentage of Fremoval does not weaken in the full pH range of 3 to 11. The adsorption of the Fions on the synthetic adsorbent is thus not dependent on pH. It is therefore not simple physisorption, and there is probably an appreciable contribution of chemisorption, resulting in a partial formation of aluminum fluoride and sodium fluoride under acidic and alkaline conditions 74 . The point of zero charge (pHPZC) plays a crucial role in determining the optimal pH of the solution where the fluoride adsorption is favored. The adsorption of these anions becomes favorable when the adsorbent surface is positively charged at a particular pH (i.e. pHsolution < pHPZC) 75 . The PZC of the aluminum-based adsorbent was measured using the powder addition method 75 , by adding 20 mL of 5 10 -2 mol.L -1 NaCl to several 50 mL cylindrical high-density polystyrene flasks. A range of initial pH (pHi) values of the NaCl solutions was adjusted from 2 to 12 by adding 10 -1 mol.L -1 of HCl and NaOH. The total volume of the solution in each flask was brought to exactly 30 mL by further addition of 5 10 -2 mol.L -1 NaCl solution. The pHi values of the solutions were then accurately noted and 50 mg of the adsorbent was added to each flask, which was securely capped. The suspensions were shaken in a shaker at 298 K and allowed to equilibrate for 48 h. The suspensions were then centrifuged at 3600 rpm for 15 min and the final pH (pHf) values of the supernatant liquid were recorded. The value of pHPZC is the point where the curve of ∆pH (pHf -pHi) versus pHi crosses the line equal to zero 76 . The obtained results are presented in Figure 8. These results show clearly that the PZC of aluminum-based adsorbent is outside a pH range of 2 to 12 because there is no intersection of the curve with the x-axis. Therefore, it can be concluded that the pH does not affect the adsorption phenomena by the aluminum-based adsorbent. This result confirms our experimental results and the pH will be no change the adsorbent capacity of the aluminum-based adsorbent.

Kinetic studies
Several models are exploited to fit the kinetic sorption tests 73 . In this study, the two models (pseudo-first-order and pseudo-second-order) were used to investigate the kinetics of the Fadsorption onto aluminum-base adsorbent, in order to understand the mechanism and the rate-controlling steps affecting the adsorption kinetics. As shown in Figure 9 and 10, the pseudo-first-order ( Figure 9) has a very low correlation coefficient (R 2 =0.7681). Contrariwise, the pseudo-second-order ( Figure 10) was found to give the best fit (R 2 =0.9911)

Adsorption isotherm
The adsorption capacity Qe (mg g -1 ) of adsorbent was examined by determining the equilibrium sorption of Fas a function of residual Fconcentration present in the liquid phase. The variation of the adsorption capacity of adsorbent for Fis presented in Figure 11. According to the equilibrium curve, the adsorption capacity at equilibrium increases progressively at lower Fconcentration. This is because of the availability of excess adsorption sites. On the other hand, as the Fconcentration increases, the adsorption capacity at equilibrium progressively decreases until reaching saturation. The availability of adsorption sites at high Fconcentration becomes the limiting factor as the adsorbent surface reaches maximum adsorption capacity. This latter was observed to be 43.29 mg g -1 .
The value of the maximum adsorption capacity obtained for this study in comparison with those reported earlier for adsorption of Fonto various adsorbents (Table 2)

Aluminum-based adsorbent 43.29 This study
The equilibrium data were further processed using Langmuir and Freundlich isotherms. The Langmuir model is based on the hypothesis that uptake occurs on a homogenous surface by monolayer adsorption without interaction between adsorbed molecules. While the Freundlich model proposes multilayer adsorption with a heterogeneous energetic distribution of active sites and with the interaction between adsorbed molecules. The Langmuir and Freundlich adsorption isotherms for the Fadsorption are presented in Figure 12.
Langmuir and Freundlich adsorption isotherm constants are indicated in Table 3.
Where C0 (mg L -1 ) is the initial concentration of Fand KL (L mg -1 ) is the Langmuir constant related to the energy of adsorption. The value of RL indicates the shape of the isotherms to be either unfavourable (RL > 1), linear (RL = 1), favourable (0 < RL <1) or irreversible (RL = 0). The calculated RL values as different initial fluoride concentrations are shown in Figure 13. It was observed that the value of RL in the range 0-1 confirms the favourable adsorption. The favourable adsorption of the Freundlich isotherm can be characterised such that if a value for n is above unity, adsorption is favourable and a physical process. In this study, the value of n (n = 2.046) is greater than 1, indicating that the adsorption is favourable. However, the value of correlation coefficient (R 2 = 0.8289) is lower than the Langmuir isotherm value (0.9837). Therefore, the Fadsorption onto aluminium-based adsorbent were better described by Langmuir isotherm confirming monolayer adsorption without interaction between adsorbed molecules.

Conclusion
The defluoridation efficiency of aluminum-based adsorbent was evaluated in a batch adsorption reactor at room temperature (25±2°C). The results of the batch adsorption experiments demonstrated that 24 h of contact time between the adsorbent and the fluoride solution was sufficient to achieve equilibrium. It was observed that the maximum wastewater (84.91%) was obtained for 40 mg L -1 and 3 g L -1 of initial Fconcentration and adsorbent dose, respectively. It is worth noting that the percentage of Fremoval does not weaken in the wide pH range of 3 to 11. The adsorption of the Fions on the synthetic adsorbent is thus not dependent on pH. Kinetic studies revealed that Fadsorption fitted well to pseudo-second-order model indicating the fluoride adsorption mechanism of aluminum-based adsorbent was of chemisorption type.
Furthermore, the adsorption isotherm of Fsorption indicated that the maximum adsorption capacity of this latter was noted to be 43.29 mg g -1 . The experimental data indicated that the Langmuir isotherm was a suitable model for describing the Fadsorption because of its high correlation coefficient.
According to the Langmuir concept, the adsorption of Fonto aluminium-based adsorbent was monolayer adsorption with homogenous distribution of active sites and without interaction between adsorbed molecules. The obtained results indicated that the ion exchange is probably the primary mechanism involved. This study indicated that aluminum-based adsorbent is an efficient material for the removal of fluoride from aqueous solutions, which can be used in water treatment without any additional modification. However, its widespread use is restricted due to high cost. Thus, the search for natural materials such as clays can be an up-andcoming alternative.

Nomenclature
C0 Initial Fconcentrations (mg L -1 ) Ce Equilibrium concentrations of F -(mg L -1 ) Kad Rate constant (min -1 ) KF Freundlich constants related to adsorption capacity (mg g -1 ) K1 Pseudo-first order rate constant (min -1 ) K2 Pseudo-second order rate constant (g mg -1 min -1 ) n Adsorption intensity (heterogeneity factor) Qt Amount of Fadsorbed per unit mass of adsorbent at time t Qe Equilibrium adsorbate capacity (mg.g -1 ) V Volume of the aqueous solution (mL) W Mass of adsorbent (g) RL Equilibrium parameter