Application of cryptocrystalline magnesite-bentonite clay hybrid for de ﬂ uoridation of underground water resources: implication for point of use treatment

A new synthesis method was established to fabricate a nanocomposite material comprising of cryptocrystalline magnesite and bentonite clay that has high adsorption capacity for ionic pollutants. To synthesizethecompositeat1:1weight(g):weight(g)ratio,avibratoryballmillwasused.Batchadsorption experiments were carried out to determine optimum conditions for ﬂ uoride adsorption. Parameters optimized included: time, dosage, concentration and pH. Optimum conditions for de ﬂ uoridation were found to be 30 min of agitation, 0.5 g of dosage, 0.5:100 solid to liquid (S/L) ratios and 25 mg L (cid:1) 1 of initial ﬂ uoride ions. Fluoride removal was independent of pH. The adsorption kinetics and isotherms were well ﬁ tted by pseudo-second-order and Langmuir models, respectively, indicating chemical and monolayer adsorption. Findings illustrated that the newly synthesized adsorbent was a promising adsorbent for the environmental pollution clean-up of excess ﬂ uoride in underground water and it can be used as a point source treatment technology in rural areas of South Africa and other developing countries.


INTRODUCTION
Considering the dangers that come with consuming high fluoride waters, a maximum permissible limit of 1.5 mg L À1 has been recommended for consumption in potable water (Ayoob &  However, adsorption is one of the techniques, which is comparatively more useful and economical at low pollutant concentration (Onyango & Matsuda ; Zulfiqar et al. ). To that end, it is the defluoridation method that is now widely used in rural areas (Masindi et al. c

Sampling
Fluoride-rich water collected from a community borehole in Siloam village Limpopo Province, South Africa was used to evaluate the effectiveness of the prepared composite adsorbent in the treatment of field water. Raw magnesite rocks from the Folovhodwe Magnesite Mine in Limpopo Province, South Africa, were collected without any prior processing. Bentonite clay was supplied by ECCA Holdings (Pty) Ltd Cape Bentonite mine (Cape Town, South Africa).

Adsorbent preparation
To prepare the material for adsorption processes, the magnesite rock samples were milled into a powder using a Retsch RS 200 mill and afterwards passed through a sieve to obtain 32 μm particle sizes. The raw bentonite was washed by soaking in ultra-pure water for 10 min and thereafter draining the wash. The level of the ultrapure water used was such that it covered the entire sample in the beaker and was allowed to overflow. The procedure was repeated four times. The washed bentonite was dried in an oven (24 h at 105 W C). The dried samples were milled into a fine powder (Retsch RS 200 mill) and sieved (32 μm particle size sieves).

Synthesis of mechanochemical activated composite
The composite adsorbent was synthesized via a mechanochemical method using the following procedure. A vibratory ball mill was used for making the magnesite-bentonite clay composite. Powdered bentonite (500 g) and magnesite (500 g) were mixed on a 1:1 wt% mass ratio.

Preparation of working solution
For all the experiments, the accuracy of the analysis was monitored by analysis of National Institute of Standards and Technology (NIST) water standards. Simulated fluoride-rich water was synthesized using sodium fluoride salt.
A standard stock solution of fluoride (1,000 mg L À1 F) was prepared by dissolving 0.221 g sodium fluoride into 100 mL deionized water from a Milli-Q water system. Fluoride-bearing solutions were prepared by diluting the stock solution to desired concentrations with ultra-pure water. Before fluoride determination, a total ionic strength adjusting buffer (TISAB III) was added to the solutions in a ratio of 10:1 in order to maintain ionic strength and pH, and eliminate the interference effect of F-ion complexing with metal cations.

Microstructural characterisation
Mineralogical composition of the composite and resulting solid residues was determined using X-ray diffraction (XRD). They were analysed using a PANalytical X'Pert Pro powder diffractometer in θ-θ configuration with an X'Celerator detector and variable divergence, and fixed receiving slits with Fe filtered Co-Kα radiation (λ ¼ 1.789 Å). The phases were identified using X'Pert Highscore plus software at University of Pretoria, South Africa. Morphology was determined using scanning electron microscopy-electron dispersion spectrometry (SEM-EDS) (JOEL JSM -840, Hitachi, Tokyo, Japan).

Optimization of adsorption conditions
Optimization experiments were done in batch experimental procedures. Parameters optimized include time, dosage, concentration and pH. All experiments were done in triplicate.
Samples of 100 mL of 50 mg L À1 F À solution were pipetted into nine, 250 mL high-density polyethylene plastic bottles and 1 g of the composite added. The mixtures were agitated for varying contact times. The mixtures were then filtered through a 0.45 μm pore nitrate cellulose filter membrane.
pH, electrical conductivity (EC) and TDS were measured using a CRISOM MM40 multimeter probe. The samples were refrigerated at 4 W C until analysis by fluoride meter.
Eight, 100 mL solutions of 50 mg L À1 F À were pipetted into eight, 250 mL bottles and varying masses of the composite added. The mixtures were agitated for an optimum time of 30 min at 250 rpm using the Stuart reciprocating shaker.
The filtered samples were treated as discussed previously. Six, 100 mL solutions of 25 mg L À1 F À were pipetted into eight, 250 mL bottles with 0.5 g of the composite and the pH was adjusted from 2 to 12 using NaOH and nitric acid. The mixtures were agitated for an optimum time of 30 min at 250 rpm using the Stuart reciprocating shaker.
The filtered samples were treated as discussed previously.
The synthetic and fluoride-rich underground water were treated under optimized conditions.

Calculation of the extent of fluoride removal and adsorption capacity
The percentage removals of fluoride by the composite were computed by the expression: where C i ¼ initial concentration and C e ¼ equilibrium ion concentration, respectively.
The amounts of fluoride adsorbed by the composite were determined by the expression: where C i ¼ initial ions concentration (mg L À1 ), C e ¼ ions concentration at equilibrium (mg L À1 ), V ¼ volume of ions solution (L) and m ¼ weight of the composite in grams.

RESULTS AND DISCUSSION
Microstructural characterizations

X-ray diffraction analysis
The mineralogical composition of magnesite, bentonite clay and magnesite-bentonite clay composite are shown in Figure 1.
XRD analysis showed that magnesite consists of periclase, brucite and forsterite as the main mineral phase. The low intensity peaks indicate that the material is enriched with amorphous phases. Bentonite clay was observed to contain smectite, quartz, plagioclase, calcite and muscovite.    verifying that the material under study is magnesium carbonate. Traces of Ca and Fe were also observed to be present.
Such metals also aid in removal of fluoride from wastewaters.
The availability of Cu is due to the use of a copper grid.

Effects of dosage
The effect of dosage on the removal of fluoride from aqueous solution is shown in Figure 5.
As shown in Figure 5, the percentage removal of fluoride was observed to increase with an increase in dosage. The percentage removal of fluoride increased rapidly, as the dosage was increased from 0.1 to 0.5 g. The composite managed to remove >99% of fluoride from the aqueous solution. After which, no significant increase in adsorption was observed. High percentage removal is attributed to more surface suitable for adsorption becoming available as the dosage increases. As such, it was concluded that 0.5 g is the optimum dosage for adsorption of 10 mg L À1 of fluoride from aqueous solution and it will be used for the following experiments.

Effect of initial fluoride concentration
The effect of initial fluoride concentration on the removal of fluoride from aqueous solution is shown in Figure 6.
The uptake of fluoride by the composite was studied by varying the initial concentration of fluoride from 2 to 50 mg L À1 . As shown in Figure 6, there was a high percentage removal of fluoride at low concentrations but as the concentration of fluoride increased the percentage removal

Effect of supernatant pH
The effect of pH on the removal of fluoride from aqueous solution is shown in Figure 7.
The effect of pH on removal of fluoride from aqueous solution was evaluated from pH 2 to 12. Removal of fluoride by the composite at varying pH was observed to be high over a wide range of pH values (Figure 7). From pH 2 to 12, the removal efficiency was greater than 99%. High adsorption of  the composite may be attributed to a high pH at point of zero charge (pH pzc ) of 10. As such at pH <10, the system is removing anions from the aqueous solution as the adsorbent is positively charged attracting anions. On the basis of this, pH of 2-10 was taken to be the optimum pH range for the subsequent experiments.

Adsorption kinetics
The effect of contact time on removal of fluoride from aqueous solution was evaluated using different kinetic models to reveal the nature of the adsorption process and the rate limiting processes. A Lagergren pseudo-first-order kinetic model is a well-known model that is used to describe mechanisms of adsorption by different adsorbents. It can be written as follows (Shou et al. ): ln q e À q t ð Þ¼ ln q e À k 1 t where k 1 (min À1 ) is the pseudo-first-order adsorption rate coefficient and q e and q t are the values of the amount adsorbed per unit mass at equilibrium and at time t, respectively. The experimental data were fitted using the pseudofirst-order kinetic model by plotting ln q e À q t ð Þ vs. t, and  the results are shown in Table 1. The pseudo-first-order was applied and it was found to fairly converge with the experimental data. Moreover, the calculated amount of fluoride ions adsorbed by the composite [q e, calc (mg g À1 )] was less than the experimental values [q e, exp (mg g À1 )] ( Table 1).
The findings indicated that the Lagergren pseudo-firstorder kinetic model is inappropriate to describe the adsorption of fluoride ions from aqueous system by the composite.
The pseudo-second-order kinetic model is another kinetic model that is widely used to describe the adsorption process from an aqueous solution. The linearized form of the pseudo-second-order rate equation is given as follows: where k 2 [g(mg min À1 )] is the pseudo-second-order adsorption rate coefficient and q e and q t are the values of the amount adsorbed per unit mass at equilibrium and at time t, respectively. An application of the pseudo-second-order rate equation for adsorption of fluoride to the composite matrices portrayed a good fit with the experimental data ( Figure 8 and Table 1). The obtained results confirm that the pseudo-second-order model is the most suitable kinetic model to describe adsorption of fluoride ions by the composite from the aqueous system. Moreover, this also confirms that the mechanism of fluoride removal from aqueous solution is chemisorption. To be noted is that the theoretical adsorption capacity is close to the experimental adsorption capacity further confirming that this model describes the adsorption data.
The overall kinetics of the adsorption process may be governed by diffusional processes as well as by the kinetics of the surface chemical reaction. In diffusion studies, the rate is often expressed in terms of the square root of time. Parameters q e, exp (mg g À1 ) q e, calc (mg g À1 ) k1 (min À1 ) R 2 Pseudo-first-order 9.9 À357 1.027 0.85 Parameters q e, exp (mg g À1 ) q e, calc (mg g À1 ) k2 (g mg À1 min À1 ) R 2 Pseudo-second-order 9.9 10 2.6 1 Parameters qe, exp (mg g À1 ) Ci (mg g À1 ) k id (mg g À1 min À1/2 ) R 2 Intra-particle-diffusion 9.9 9.0 0.17 0.7 where k id (mg g À1 min À1/2 ) is the intra-particle diffusion coefficient (slope of the plot of q t vs. t 1=2 ) and C i is the intra-particle diffusion rate constant. The results show that the intra-particle diffusion model was not applicable for the present process due to the lower correlation coefficients as shown in Table 1. Different kinetic model parameters for the adsorption of fluoride ions onto the composite are shown in Table 1.
Pseudo-second-order plot for fluoride removal by the composite is shown in Figure 8.

Adsorption isotherms
The relationship between the amount of ions adsorbed and the ion concentration remaining in solution can be described by adsorption isotherms. The two most common isotherm types for describing this type of system are Langmuir and Freundlich adsorption isotherms. These models describe adsorption processes on a homogenous (monolayer) or heterogeneous (multilayer) surface, respectively.
The most important model of monolayer adsorption came from Langmuir. This isotherm is given as follows: The essential characteristics of the Langmuir isotherms can be expressed in terms of a dimensionless constant separation factor or equilibrium parameter, R L , which is defined as: R L >1 means unfavourable reaction, R L ¼ 1 means linear fit, 0 < R L < 1 means favourable reaction and R L ¼ 0 means irreversible reaction.
Where C e ¼ equilibrium concentration (mg L À1 ), Q e ¼ amount adsorbed at equilibrium (mg g À1 ), Q m ¼ Langmuir constants related to adsorption capacity (mg g À1 ) and b ¼ Langmuir constants related to energy of adsorption (L mg À1 ). A plot of C e versus C e =Q e should be linear if the data are described by the Langmuir isotherm. The value of Q m is determined from the slope and the intercept of the plot. It is used to derive the maximum adsorption capacity and b is determined from the original equation and it represents the intensity of adsorption. The Langmuir adsorption isotherm plot is shown in Figure 9.
The Freundlich adsorption isotherm describes the heterogeneous surface energy by multilayer adsorption. The Freundlich isotherm can be formulated as follows: where C e ¼ equilibrium concentration (mg L À1 ), q e ¼ amount adsorbed at equilibrium (mg g À1 ), K ¼ partition coefficient (mg g À1 ) and n ¼ intensity of adsorption. The linear plot of log C e versus log q e indicates if the data are described by the Freundlich isotherm. The value of K implies that the energy of adsorption on a homogeneous surface is independent of surface coverage and n is an adsorption constant which reveals the rate at which adsorption is taking place.
These two constants are determined from the slope and intercept of the plot of each isotherm. The plot of Freundlich adsorption isotherm is shown in Figure 10.
The parameters of Langmuir and Freundlich adsorption isotherms are shown in Table 2.
As tabulated, the Langmuir isotherm showed a high correlation coefficient (R 2 > 0.99) (Figure 9). R L shows that the reaction of fluoride and the composite was favourable. The value between 0 and 10 shows that the reaction is beneficial and the K f value shows that it has high adsorption capacity.
This indicates a good agreement between the experimental values and isotherm parameters. The data fitted better to the Langmuir adsorption isotherm than the Freundlich adsorption isotherm thus depicting a monolayer adsorption mode.

Removal of fluoride under optimized conditions
Results for the removal of fluoride on raw water under optimized conditions are shown in Table 3. The physiochemical conditions of borehole water before and after defluoridation are also tabulated in Table 4. The fluoride-rich ground water was observed to be slightly alkaline with a pH of 9. The composite was observed to remove fluoride from groundwater to below 0.01 mg L À1 . This shows that the composite is an effective material that can be used for the removal of fluoride in groundwater.
Adsorption capacity of the composite as compared to other adsorbents A comparison of the adsorption capacity of the composite and other adsorbents that have been reported to remove fluoride is shown in Table 4.    initial fluoride concentration, pH 10 and 250 rpm shaking speed. Greater than 99% removal efficiency for fluoride was observed at these optimum conditions. The adsorption data fitted better to the Langmuir adsorption isotherm than the Freundlich adsorption isotherm therefore proving monolayer adsorption. Adsorption kinetics fitted better to pseudo-second-order than pseudo-firstorder thus indicating chemisorption. This study produced a novel engineered material with better adsorption capacity for fluoride when compared to other conventional methods.