An eco-friendly and low-cost strategy for groundwater defluorination: Adsorption of fluoride onto calcinated sludge

The excess of fluoride ions (F ) in water for human supply is a serious public health. The recommended concentration of F ions by the World Health Organization (WHO) is 1.5 mg L-1. Several groundwater sources around the world contain high F concentrations, and require treatment before human consumption. It was developed an eco-friendly and low-cost strategy for groundwater defluorination, i.e., adsorption onto calcinated sludge. This strategy was efficient at pH of 5.5 and using 5 g L-1 of calcinated sludge. The groundwater attained the WHO standard within 60 min. The kinetic model of pseudo-second-order obtained a better adjustment to the experimental data. The equilibrium curve at 25 C was better represented by the Tóth model. The maximum adsorption capacity was 2.04 mg g-1. Therefore, adsorption using calcinated sludge can be considered as an ecofriendly and low-cost strategy for groundwater defluorination.


Introduction
The water demand is increasing worldwide, and it is necessary to provide a suitable treatment to adequate it from distinct sources into potable regulation standards. From conventional treatment, great quantities of sludge are generated as solid waste, whose disposal must be adequate due to the use of iron-and aluminum-based metallic coagulants during the coagulation-flocculation stage [1,2]. In general, its composition is an amount rich in organic matter added to inorganic compounds, depending on factors as spring of raw water catchment and the applied treatment [2]. In the case of sludge containing trivalent aluminum ions, it is not possible to dispose of them on the soil or water bodies [3]. Such ions, however, confer a considerable surface area and potential for use as an adsorbent for several objectives [4].
Another serious problem is the contamination of the surface and groundwaters by the excess of fluoride ions. People exposed to the intake of such waters for a long period can suffer from fluorosis, a dental type of disease in the first stage, and skeletal when in an advanced stage [5]. The source of contamination in such water bodies can be both natural and anthropogenic, occurring in many places around the world [6]. The World Health Organization's recommendation for the maximum concentration of fluoride ions is of 1.5 mg L − 1 , which is enough to fight the dental cavity [7]. On the contrary, water treatment must be proceeded to guarantee its potability.
There are different techniques for water defluorination, being them precipitation/coagulation, ion exchange, membrane separation, electrochemical treatment, and adsorption [6,[8][9][10]. Among those, adsorption stands out under environmental and economic aspects besides its simple design and operation with good removal rates [8]. Thus, it is viable to apply this technique in remote regions supplied mainly by groundwater [11]. The most used adsorbent material is activated alumina, in line with the recommendations from the United States Environmental Protection Agency, with a good capacity of removal and regeneration, but with relatively high costs [12]. Other materials used are activated charcoal, bone char, mineral rocks, biosorbents, or residues, which present low costs and high local availability, as is the case of sludge [13].
In this work, it was investigated the potential of a calcinated sludge for groundwater defluorination. Sludge, which is a waste of great abundance and at no cost generated in conventional water treatment plants, comes specifically from the coagulation-flocculation step, which through its structural characteristics, such as chemical composition, can be used as an adsorbent for removing fluoride ions from both synthetic solutions and groundwater. The sludge was from a local water treatment plant (Santa Maria, RS, Brazil). The material was calcinated and carefully characterized. It was sampled groundwater from Faxinal do Soturno (RS, Southern Brazil) and synthetic solution of sodium fluoride for comparison purposes. The treatment was optimized by response surface methodology (RSM). Once the best conditions were obtained, kinetic and equilibrium studies were carried out. Thus, a treatment for the important public health problem, which is the excess of fluoride ions, capable of reducing the concentration under 1.5 mg L − 1 , was proposed, through the reuse of a residue continuously generated in great amounts in conventional water treatment plants.

Obtainment, treatment, and characterization of the sludge
Raw sludge was collected from a water treatment plant (Rio Grande do Sul sanitation company, CORSAN) located in Santa Maria, RS, Brazil. The material was oven-dried at 80 • C for 24 h. Then, 50 g of the material was calcinated in an oven (device F3000, EDG brand) at 200 • C for 2 h with a heating rate of 10 • C min − 1 , in an oxidizing atmosphere. The sludge was then washed with 1 L of a 20 % (v/v) alcohol/aqueous solution under magnetic stirring for 1 h. The solid material was separated by filtration and oven-dried at 80 • C for 24 h. The chemical species were acquired from Neon (Brazil) with analytical grade.

Groundwater and fluoride solutions
The groundwater employed in this work was collected in Faxinal do Soturno, RS, Brazil. The groundwater characteristics are presented in Table 1. The variation of these characteristics is due to the temperature and seasons. Four samples were taken every two months from August 2018 to June 2019. For comparison with groundwater, synthetic fluoride solutions were also used in the defluorination experiments. These solutions were prepared with deionized water and sodium fluoride (Analytical grade, Belga Química, Brazil). A stock solution of 100 mg L − 1 was prepared and diluted to obtain the concentrations of interest.
The fluoride concentrations were determined through ion chromatography (model: 930 Compact IC Flex, Metrohm, Swiss), with conductivity detector, chemical suppression, and CO 2 suppression. The compound separation was performed by an ionic change column (Metrosep A sup 5-150/4.0, Metrohm, Swiss) and pre-column (Metrosep A sup 5 Guard, Metrohm, Swiss) both with quaternary ammonium groups supported on polyvinyl alcohol, and accuracy of μg L − 1 . The sample was pumped with a flow rate of 0.7 mL min − 1 at a temperature of 30 • C. The eluent concentration presented the following composition: 3.2 mmol L − 1 of sodium carbonate (Na 2 CO 3 , 99.95 % purity, Sigma Aldrich, USA), 1.0 mmol L − 1 of sodium bicarbonate (NaHCO 3 , 99.8 % purity, Sigma Aldrich, USA) and 20 % of acetone (CH 3 COCH 3 UV/HPLC grade, Sigma Aldrich, USA). Sulfuric acid was employed as a chemical suppressor (H 2 SO 4 , 96.0 % purity, Sigma Aldrich, USA). The anion standard was prepared from a multi anions solution (bromide, chloride, nitrate, sulfate, fluoride and phosphate, SpecSol, Brazil).

Defluorination experiments
Defluorination experiments were carried in three steps: optimization experiments, kinetic experiments, and equilibrium experiments. All these experiments were carried out with plastic Erlenmeyers containing    100 mL of solutions, which were immersed in a Dubnoff type orbital shaker (MA095, Marconi) under the agitation of 150 rpm and 25 • C. For comparison, all the tests were performed with groundwater and also with synthetic fluoride solutions. After all experiments, samples were centrifuged, and the liquid was used for quantification. In the optimization experiments, solutions were stirred for 2 h. The adsorbent dosage ranged from 1 to 5 g L − 1 , and the pH ranged from 4 to 10. For the groundwater, the initial concentration of fluoride ions was 4.31 mg L − 1 , and for the tests with the synthetic solutions, the initial concentration was 5.37 mg L − 1 . In the kinetic experiments, curves were constructed with groundwater (initial F-concentration of 3.60 mg L − 1 ) and synthetic solutions (initial F-concentration of 5.37 mg L − 1 ). The curves were obtained in different pH values, i.e., 8.83 (natural pH) and 5.5. The adsorbent dosage was 5 g L − 1 , and aliquots were withdrawn at 2.5, 5, 7.5, 10, 15, 20, 30, 60, 120, 240, and 300 min. For the isotherm experiments, curves were obtained with pH and initial F-concentration for groundwater and synthetic solutions equal to the kinetic tests. The adsorbent dosage varied from 1 to 15 g L − 1 , and the solutions were stirred until the equilibrium.

Optimization of the defluorination process
Defluorination of groundwater and synthetic solutions using calcined sludge was optimized according to the adsorbent dosage and pH. The pH and dosage are operational conditions that influence the adsorption process/defluorination. The pH of the medium controls the magnitude of the electrostatic charges that are transmitted by the ionized adsorbate molecules, thus directly affecting the number of molecules that can be adsorbed. Also, the relation between the point of the zero charge (pH pzc ) and the pH can further increase the removal of fluoride. Concerning the dosage, the effects are related to the number of available sites per gram of material. In other words, the increase of the dosage causes an increase in the percentage of removal. However, higher dosages lead to lower adsorption capacity. Thus, aiming the optimization of these two independent variables, a Central Composite Rotational Design (CCRD) was used, this design considers 2 independent variables (k), and coded levels of -α, -1, 0, + 1, +α and axial spacing of 1.41 (2k/4). The levels and variables of CCRD are depicted in Table 2. The considered responses were fluoride removal percentage (R, %) and adsorption capacity (q e , mg g − 1 ). The results were analyzed with the aid of the software Statistica 8.0 (StatSoft, USA), proceeding to the analysis of variance (ANOVA) and the verification of the determination coefficient (R 2 ) and Fischer test to evaluate the variance and prediction of the model [17].

Kinetic and equilibrium of defluorination
The kinetic data describing the adsorption capacity of the calcinated sludge for fluoride (q t , mg g − 1 ) over time t (min) were adjusted with several reaction models presented in the literature [18][19][20][21]. The models  The equilibrium data describing the adsorption capacity of the calcinated sludge for fluoride (q e , mg g − 1 ) versus the remaining fluoride concentration in the solution (C e , mg L − 1 ) were fitted with some isotherm models presented in the literature [22][23][24][25]. The models and their respective details are presented in the supplementary file (S.3.). The fit quality of the kinetic and isotherm models was evaluated by the determination coefficient (R 2 ), adjusted determination coefficient (R 2 adj ), average relative error (ARE), and minimum squared error (MSE) (Supplementary file, S.4.). Fig. 1 shows the XRD spectra of the calcinated sludge used as a fluoride ion adsorbent. The sludge possesses a semi-crystalline structure, with a predominance of amorphous regions, according to the wide reflection range from 18 to 40 • . The sludge has a complex chemical composition, containing predominantly oxides of silicon aluminum and iron, a structural characteristic that is shown in the diverse peaks found in the XRD. The peak at 20.73 • and 26.5 • indicates the presence of quartz in the calcined sludge [26,27]. The peak at 24.75 • may be due to the presence of TiO 2 [28]. The peak at 37 • may be due to the presence of Al 2 O 3 and Fe 2 O 3 in the calcined sludge [28]. The peak at 41.67 • can be attributed to the presence of iron oxide in the calcined sludge [29]. The 50 • peak can be attributed to the presence of quartz in the sample [30]. The peak at 68 • may be due to the presence of Al 2 O 3 in the calcined sludge [31].

Calcined sludge characteristics
The Fourier transform infrared spectrum of the calcinated sludge ( Fig. 2) presents characteristic bands of materials of the same nature as the 3400− 3700 cm − 1 band associated with the stretching of the OH bonds [32]. The remaining organic matter presents several characteristic bonds in the range of 1050− 1725 cm − 1 , which may be C--C aromatic and C--O carbonyl conjugated in 1635-1643 cm − 1 [33], vibrations due to the H-O-H bond between 1650− 1600 cm − 1 [32], OH deformation, symmetric COOstretching and C -O stretching of phenolic groups in 1400− 1465 cm − 1 [34]. The Si-O bond is found at     [11,[35][36][37]. Besides, according to [38], bands of 912 and 539 cm − 1 are associated with Al-O bonds that suggest the presence of amorphous forms of aluminum oxide and hydroxide (alumina structure), which favors the adsorption of fluoride ions. Scanning electron microscopy of the calcinated sludge is shown in Fig. 3. The calcinated sludge presents a regular surface and a granular structure of varying sizes, well distributed, with characteristic roughness. The atomic percentages of the EDS analysis in Table 3 indicate the predominance of Fe, Al, Si, and O, components found in the sludge in the form of oxides (SiO 2 , Al2O 3, and Fe2O 3 ), present due to the use of iron and aluminum-based metallic coagulants in the coagulation-flocculation stage in water treatment plants [32].
Figs. 4 (a) and (b) show the adsorption/desorption isotherm of N 2 and the pore size distribution of the calcinated sludge, respectively. Fig. 4 (a) shows a type IV isotherm, where N 2 adsorption occurs through multiple layers adsorption, followed by capillary condensation. The hysteresis loop between P/P 0 of 0.6− 0.9 indicates the predominance of mesopores in the sludge structure [39]. Fig. 4 (b) shows that the largest porous fraction is in the mesoporous region (which comprises the regions from 2 to 50 nm). The results of the BET analysis showed that the surface area of the calcinated sludge was 70.1 m 2 g − 1 ; the average pore size was 11.3 nm, and the average pore volume was 0.12 m 3 g − 1 .
The point of zero charge (pH PZC ) is the pH value of the medium in which the adsorbent material has zero surface charge. In this work, the value obtained for the calcinated sludge was 5.6. Consequently, at pH values below 5.6, the surface of the calcinated sludge is positively charged and, therefore, there is a greater attraction for species with negative charges, such as fluoride [40,41]. Otherwise, at pH greater than 5.6, the surface is negatively charged, resulting in repulsion between the adsorbent particles and the fluoride ions. Fig. 5 shows the thermogravimetric profile of the calcined sludge. The increase in temperature under an oxidizing atmosphere causes significant changes in the sample mass, which can be identified through the thermogravimetric analysis curve (Fig. 5). The first peak observed at 155 • C is associated with the loss of free or physically bound water to the sludge [27]. Remnants of moisture are still eliminated between 155 and 400 • C, a temperature range that is also attributed to the oxidation of  organic matter present in the sludge. However, a part may still be retained on its surface [42]. This possibly explains the resulting adsorption capacity and the characteristic functional groups identified in infrared spectroscopy, in addition to the slight loss of mass found for temperatures up to 300 • C. The second peak that starts at approximately 375 • C is exothermic and represents the loss of water contained in the aluminum hydroxide and its phase change to γ-Al 2 O 3 , also called dehydroxylation [26,31]. This process can also occur with clay materials [27,43]. Thus, the material matrix ends up being highly modified for temperatures above 500 • C, which can compromise its fluoride ion adsorption capacity.
It should be considered that the physical and chemical characteristics of the adsorbent plays a major role in the adsorption/removal of fluoride ions. In this case, the characterization techniques show that the that sludge is a complex mixture of several metal oxides, such as aluminum and iron oxides. Besides that, the structure of the calcinated sludge is mainly mesoporous, with a specific surface area of 70 m 2 g − 1 . The conjunction of these characteristics and the composition of the sludge can improve and directly influence the performance of the adsorbent.

Optimal conditions for defluorination using calcinated sludge
Defluorination using calcinated sludge was optimized by a central composite rotatable design (CCRD) type (2 2 ). The optimization was performed in terms of pH and adsorbent dosage. The considered responses were fluoride removal percentage (R, %) and adsorption capacity (q e , mg g − 1 ). The process was optimized for both a synthetic fluoride solution and a real sample of groundwater collected from a well in Faxinal do Soturno, RS (Brazil). Table 4 shows the matrix used in the experimental design and the respective results of R and q e , for the adsorption of fluoride ions from groundwater and synthetic solution. Table 4 shows that lower pH values associated with high adsorbent dosages provide greater removals of fluoride ions. However, the adsorption capacity is negatively affected by the increase in adsorbent dosage for both systems.
Pareto graphs for fluoride adsorption from groundwater are shown in Fig. 6 (a) and (b). Pareto graphs for fluoride adsorption from synthetic solutions are depicted in Fig. 6 (c) and (d). From these graphs, it was possible to determine the significant variables for each response, removal (R), and adsorption capacity (q e ). In the graphs, positive values indicate that the independent variables are directly proportional to the response, and negative values show that the independent variables are inversely proportional to the dependent variables.
Taking into account the fluoride adsorption from groundwater, according to Fig. 6 (a), only the interaction between the two variables was not significant in a 95 % confidence interval for the response removal percentage. In Fig. 7 (b), only the quadratic adsorbent dosage variable was not significant within a 95 % confidence interval for the response adsorption capacity. Thus, excluding the non-significant variables, the equations that describe the R and q e responses as a function of pH and adsorbent dosage are Eq. (1) and Eq. (2): Where X 1 represents the pH, and X 2 represents the adsorbent dosage.  The determination coefficients (R 2 ) for Eqs. (1) and (2) were 0.95 and 0.81, respectively, and indicate a satisfactory adjustment of the experimental data to the proposed model.
Considering now the fluoride adsorption from synthetic solutions, for the response R (Fig. 6 (c)), only the interaction between the two variables was not significant, therefore excluding this variable, the model can be given by Eq. (3). For the adsorption capacity response (q e ), according to Fig. 6 (d), all variables were significant so that the model can be given by Eq. (4): Where X 1 represents the pH, and X 2 represents the adsorbent dosage. The determination coefficients (R 2 ) for Eqs. (3) and (4) were 0.96 and 0.97, respectively, and indicate an optimal adjustment of the experimental data to the proposed model.
Thus, to generate the surface curves that represent R and q e as a function of pH and adsorbent dosage, the prediction of the models was evaluated using the Fisher test. For groundwater, the value of tabulated F for Eq. (1) Fig. 7 (a) and (b) show the surface curves for fluoride adsorption from groundwater, for the responses R and q e , respectively. In both figures, a decrease in the pH value leads to greater removal of fluoride ions and greater adsorption capacity. This trend can be explained by the point of zero charge of the calcinated sludge, which is 5.6. In this way, at lower pH values, the sludge surface is positively charged, favoring the electrostatic attraction of negative ions, like fluoride. The adsorption capacity response (q e ) was inversely influenced by the adsorbent dosage, since. However, taking into account the removal percentage, higher adsorbent dosages lead to greater fluoride removals, since a greater number of active adsorptive sites are available. It should be highlighted in the tests with groundwater that a removal above 65 % indicates that the concentration reaches the limit of fluoride ion concentration permitted by the World Health Organization. Fig. 8 (a) and (b) show the surface curves of the fluoride ion removal (R) and adsorption capacity (q e ) for the experiments using the synthetic solution. Again, both for fluoride ions removal and adsorption capacity, a decrease in the pH value leads to an improvement in the response values, with pHs below 7 already resulting in removals greater than 70 %, ensuring the reach of limits established by World Health Organization. The adsorption capacity was negatively affected by the increase of the adsorbent dosage, having maximum adsorption capacities of 2.07 mg g − 1 at a pH of 4.9 and an adsorbent dosage of 1.58 g L − 1 .
Taking into account the recommendation of the World Health Organization, (limit of 1.5 mg L − 1 of fluoride ions present in potable water), the experiments to determine the kinetics and adsorption isotherms were carried out with the highest adsorbent dosage used in the experimental planning, so that such requirements was met. For the pH, it was observed that lower values lead to greater removals, but the pH  chosen for performing the kinetics and isotherms was 5.5. This value is lower than the calcinated sludge's point of zero charge (5.6), ensuring that the surface of the material is positively charged, since the surface curves show that good removals were achieved in that pH range. Besides, performing kinetic and isothermal experiments at a pH not as acidic as 4 guarantees a lesser deviation from reality.

Adsorption kinetics of fluoride ions on calcinated sludge
Kinetics curves were constructed for the fluoride adsorption on calcinated sludge from groundwater and synthetic solutions at different pH values (8.8 and 5.5) (Fig. 9). The pseudo-second-order model was the most adequate to represent the adsorption kinetics of all systems, as shown in Table 5 and Fig. 9. This occurred due to the high values found for the determination coefficient (R 2 ), adjusted determination coefficient (R 2 adj ), low values of the relative average error (ARE), and MSE. The pseudo-second-order kinetic model assumes that that adsorption occurs due to the participation of valence forces or electron exchange between the adsorbate/adsorbent, which is in agreement with the attraction of fluoride ions and adsorbent surface.
The term q 2 , which represents the theoretical adsorption capacity of the pseudo-second-order model, was higher when the adsorption experiments occurred at the lower pH. This trend corroborates with the results found for the point of zero charge. The maximum theoretical capacity found for the groundwater sample at pH 5.5 was 0.64 mg g − 1 , while the synthetic solution, at the same pH, was 1.05 mg g − 1 . This difference is mainly due to the presence of other ions in the groundwater sample such as chloride, bromide and sulfate (Table 1), which competes with fluoride for the sorption sites. The theoretical values close to the experimental values also indicate an optimal adjustment of the data to the model.
The kinetic curves of both groundwater and synthetic solution tend to a constant adsorption capacity value in a relatively short time of 240 min. For the groundwater sample at a pH of 5.5, a time of just 30 min already causes a 60 % removal in the initial concentration of fluoride  ions, going from 3.6 mg L − 1 to 1.49 mg L − 1 , reaching the standards set by the World Health Organization. For a more concentrated synthetic solution, with an initial concentration of 5.37 mg L − 1 , it takes 60 min to reach the limit stipulated by the World Health Organization. Thus, the calcinated sludge allows, in a short period, to make the water contaminated with fluoride ions drinkable for human consumption and within the WHO standard.

Adsorption isotherms of fluoride ions on calcinated sludge
Isotherm curves were constructed for the fluoride adsorption on calcinated sludge from groundwater and synthetic solutions at different pH values (8.8 and 5.5). The adsorption isotherm data are shown in Fig. 10 and Table 6. Considering the high values of the determination coefficient (R 2 ) and the adjusted determination coefficient (R 2 adj ), and the low values of the average relative error (ARE) and MSE, the model that best described the adsorption isotherms was the Tóth model. This model is a variation of the Langmuir model, applicable to adsorption on heterogeneous surfaces, where most sites have adsorption energy less than the maximum. The fit of this model is following the heterogeneous characteristics of the calcinated sludge used in this research.
The lower pH values lead to a greater adsorption capacity of the calcinated sludge, where the maximum theoretical capacity found in groundwater was 2.04 mg g − 1 , whereas, for the synthetic solution, the maximum capacity found was 3.18 mg g − 1 . At pH 8.8, due to the surface charge presented by the calcinated sludge, lower values of theoretical adsorption capacity were found, of 0.61 mg g − 1 in groundwater and 1.05 mg g − 1 in synthetic solution. Table 7 shows a comparison of several adsorbents for the adsorption of fluoride ions. It is noticed that the calcinated sludge has good adsorption capacity compared to the other adsorbents. Besides, the calcinated sludge used in this work has advantages regarding the other adsorbents, such as simplicity for preparation, ease, abundance, and the cost of obtainment. These findings make the calcinated sludge from water treatment stations a potential adsorbent for the removal of fluoride ions from groundwater.

Conclusion
In the present study, it was proposed an eco-friendly and low-cost strategy for groundwater defluorination. Sludge from a water treatment station was modified through calcination and applied as an adsorbent material in the removal of fluoride ions from groundwater. Calcinated sludge presented a semi-crystalline and mesoporous structure, with a specific surface area of 70.1 m 2 g − 1 , an average pore size of 11.3 nm, and an average pore volume of 0.12 m 3 g − 1 . The functional groups and other components of the raw sample remained, especially those based on aluminum, iron, and silica, with the elimination of only part of the organic matter due to the calcination process.
The optimal condition for groundwater defluorination was a pH of 5.5 and an adsorbent dosage of 5 g L − 1 . Under these conditions, the fluoride removal was higher than 70 %; the adsorption capacity was satisfactory and, the WHO limits for fluoride were attained. The kinetic model of pseudo-second-order obtained better adjustment to the experimental data, where times of just 60 min resulted in groundwater with fluoride concentrations below 1.5 mg L − 1 . For the equilibrium curves, the Tóth model was the one that obtained the best adjustment with the experimental data. The maximum adsorption capacity was 2.04 and 3.18 mg g − 1 for the groundwater sample and synthetic solution, respectively. It was also observed that the ions present in groundwater, including chloride, bromide, and phosphate, negatively affect the fluoride adsorption capacity. From the results presented, it is possible to consider calcinated sludge from a water treatment station as a potential adsorbent for the removal of fluoride ions from groundwater.

Declaration of Competing Interest
The authors report no declarations of interest.