Study the Removal of Fluoride from Aqueous Medium by Using NanoComposites

Endemic fluorosis disease has become a major geo-environmental health care issue caused by fluoride ion. High-efficiency and low-cost materials to uptake fluoride from water have been a challenge for scientists and engineers. Here, we report a low-cost process by utilising low-cost starting materials to develop nanocomposite adsorbents for fluoride uptake from water. Bermuda grass as a starting source material converted into nanocomposite carbon fibers upon heat treatment at 800 ̊C for one hour in Nitrogen atmosphere in the presence of metal oxides. Iron oxide-based nanocomposite (IBNC) is performing high (≈97%) removal of fluoride ion at a contact time of 60 minutes (pH 4) followed by titania-based nanocomposite (TBNC) (≈92%) and micro carbon fiber (≈88%) respectively. The phenomenon of fluoride ion uptake is realised by Freundlich adsorption model, and both adsorption capacity and adsorption intensity for IBNC are higher than those for TBNC and micro carbon fiber.


Introduction
Over the years groundwater has been regarded as the safest and most protected source of water, fit for drinking without treatment.There is a modest consideration given to the risks of chemical pollution, particularly to the presence of elevated levels of fluoride, arsenic and nitrate in groundwater.Recent studies, however, show the alarming threat of groundwater contaminations and the urgent need to find a low-cost treatment process for hazardous elements prior to drinking in different regions of the world.Specifically in the case of fluoride ions, con-sumption of water having excess fluoride over a prolonged period leads to a chronic disease known as fluorosis [1]- [4].Fluoride is normally present in the form of fluoride ions in minerals, found in sedimentary (limestone and sandstone) and igneous (granite) rocks, and leached out to contaminate water through geological processes.Higher fluoride levels in the groundwater get through these minerals by volcanic and fumarolic processes.Dissolution of these barely soluble minerals depends on the water composition and the time of contact between the source minerals and the water.Fluoride in drinking water may be beneficial or detrimental depending on its concentration [5].The upper limit of fluoride concentration in the natural drinking water is 1.5 mg/L according to the guidelines for drinking water quality [6]- [8].
Natural water is becoming enriched with fluoride by the means of (a) geological processes, and (b) industry.Water contamination by industry stems from many areas including: effluent discharge, fertilizers & pesticides, fluorosilicone and fluorocarbon polymer synthesis, coke manufacturing, glass and ceramic manufacturing, electronics manufacturing, electroplating operations, steel and aluminum manufacturing, metal etching (with hydrofluoric acid), and wood preservatives.Average fluoride values for aluminium reduction plants are reported as 107 -145 mg/L in waste water streams [4].Concentrations of an order of magnitude greater have been reported for glass manufacturing, ranging from 1000 to 3000 mg/L of fluoride [5] [6].Fluoride wastes are also generated in the steel manufacturing industry from the sintering and steel-making processes [7]- [12].In the sintering plant fluoride is found as an impurity of limestone, where both fluorspar (CaF 2 ) and limestone are among the basic materials used in steel making and as such, contributed as the principal source for this industry of fluoride containing wastewater.
Incidences of high fluoride groundwater have been reported from 23 nations around the globe [13].Endemic fluorosis has become a major geo-environmental health issue in many developing countries of the world.The Bureau of Indian Standards outlined Indian standards stating that 1 mg/L was the maximum permissible limit of fluoride with further remarks stating "the lesser the better" [14].Of the fluoride deposits of 85 million tons on the earth's crust, 12 million tons are found in India [15].Seventeen states in India are endemic for fluorosis [16] [17].About 25 million people in India suffer from fluorosis and another 66 million are at risk [18].It was stated in 1993 that crippling skeletal fluorosis might occur in people who have ingested fluoride of 10 -20 mg per day over 10 to 20 years.Early stages of skeletal fluorosis start with pain in bones and joints, muscle weakness, stiffness of joints, and chronic fatigue.During later stages of fluorosis, calcification of the bones takes place; osteoporosis in long bones and symptoms of osteosclerosis appear where the bones become denser and develop abnormal crystalline structure [19].General skeletal fluorosis directly affects the economy of villagers (mostly tribal populations) as it causes illness and debilitation not only in humans but also in their domestic animals, on which they depend for their basic income [20].Asian countries mainly India and China are worst affected.In 1993, 5 million people in Mexico (about 6% of the population) were affected by fluoride in groundwater [21].Fluorosis is prevalent in some parts of central and western China, caused not only by drinking fluoride that is in the groundwater but also by breathing airborne fluoride released from the burning of fluoride-laden coal.Worldwide such instances of industrial fluorosis are on the rise.Argentina in Latin America and East and North Africa are also endemic.Dental fluorosis is also caused due to damage of tooth enamel by the fluoride ion.Tooth enamel is made up of hydroxyapatite (87%) which is a crystalline calcium phosphate [22]; the hydroxide ion of this enamel is displaced by the most stable fluorideion to form fluoroapatite, a trigger of dental fluorosis.Humans are most prone to this during the first 7 years of life.Dental fluorosis has been characterized and reported when fluoride ion intake levels rise above 1.5 mg/L for an extended period of time.
Several different materials have been used for defluoridation such as activated carbon (AC), tricalcium phosphate, synthetic ion exchangers, lime activated alumina, and alum.There have also been many methods developed to remove excessive fluoride from water including adsorption, ion exchange, precipitation, electrolysis, donnan dialysis, and electro dialysis [23]- [43].Among these methods, adsorption is still one of the most extensively used and cheapest methods for defluoridation of water [44]- [51].However, in recent years, a considerable amount of attention from scientists has been devoted to the study of different types of low-cost materials for fluoride uptake such as: spent bleaching earth, wollastonite and chine clay, bentonite and activated bentonite, kaolinitic clay, agricultural by-products, fly ash, carbon slurry, biogas residual slurry, zeolite, bone char, and flax shive [52]- [56].Literature research has shown that alumina supported on carbon nanotubes (Al 2 O 3 /CNT) has a saturation adsorption capacity of 39.4 mg/g fluoride uptake.It is also found that the adsorption capacity of Al 2 O 3 /CNT is 3.0 -4.5 times that of γ-Al 2 O 3 while almost equal to that of IRA-410 polymeric resin at 25˚C.The adsorption isotherms of fluoride on Al 2 O 3 /CNT fit the Freundlich equation well with the optimal pH ranging from 5.0 to 9.0.Crystalline and hydrous iron (III)-zirconium (IV) hybrid oxide (IZHO) at the optimum pH range between 4.0 and 7.0 provides good efficiency of fluoride uptake [29] [57].The kinetic data obtained for fluoride adsorption on IZHO at pH 6.8 (±0.1) and room temperature (303 ± 2 K) described that both the pseudo-firstorder and the reversible first-order equations are equally well (r 2 = ~0.98 -0.99), and better than the pseudosecond-order equation (r 2 = ~0.96-0.98) for higher concentrations (12.5 and 25.0 mg/dm 3 ) of fluoride uptake.The synthesis of a zirconia/multi-walled carbon nanotube (ZrO 2 /MWCNTs) hybrid is also a novel sorbent for water defluoridation.The ZrO 2 /MWCNTs were significantly more effective in fluoride removal than all other sorbents tested before, including ZrO 2 impregnated carbons.The kinetics of fluoride adsorption on the mixed oxide is driven by boundary layer diffusion [58]- [60].
Recently, many researchers have worked on cost-effective fluoride adsorbents such as zeolites, metal-polymer & metal-metal oxide composites as well as biomass materials like fishbone charcoal, Cynodon dactylon, chitosan, acacia Arabica plant, and Tamarindus indica.Mostly, biosorption adsorbents have shown the fluoride removal to be no better than 88%, though this is still better than a metal oxide system.It has been proven that biosorption has good affinity selectively for ion uptake from aqueous solutions [61]- [65].
The present work focuses on a hybrid system for fluoride uptake where a low-cost biosorbent, using Bermuda grass (i.e., Cynodon dactylon) that is specially processed, leads to a novel fibrous material that once obtained and impregnated with titania and iron oxide in-situ by thermal chemical vapour deposition (thermal-CVD) technique, is ready for fluoride uptake.

Processing of Bermuda Grass
Bermuda grass is a natural bio fiber that is abundantly available in India which was specially processed for our investigation.The cleaned dry grass was pulverized and the fibrous powdered material was washed well with doubly distilled water to remove the free acid, followed by thermal treatment using chemical vapour deposition (CVD) setup at a heating rate of 50˚C/min in the presence of flowing N 2 (250 mL/min).The heating continued up to a final temperature of 600˚C, where it was held for 5 minutes, to obtain thermally activated carbon without using any catalysts.The resulting product was cooled to room temperature and sieved to the desired particle sizes, namely 200 -325 mesh (74 to 44 µm).Finally, the product was stored in vacuum desiccators until required for further experiments.

Conversion of Micro Carbon Fiber to Nano Carbon Fiber Using Nano-Oxides
The next set of experiments was conducted using catalysts such as TiO 2 and Fe 2 O 3 nano powders.Titania was procured from Aldrich chemical supplier whereas nanoiron oxide is plentiful available in our mill as waste by product of hot rolling steel.These catalysts were mixed with pulverized raw grass powder in the ratio of 1:100 (catalyst to grass powder) and fed into the CVD set up at a heating rate of 50˚C/min in the presence of N 2 flow (250 mL/min).The dwell time of the mixed samples was 1 h each and the rate of cooling was the same as that of heating.We could ascertain the optimum temperature for obtaining Fe 2 O 3 based carbon nanofiber and TiO 2 based carbon nanofiber is 800˚C from their respective phase diagrams for carbide formation.

Fluoride Adsorption Study
In the present study, the derived nanocomposites were utilized for the removal of fluoride from its aqueous solution.The sorption isotherm and kinetics experiments were performed by batch adsorption experiments and were carried out by mixing 0.5 g (obtained by the study of effect of adsorbent dose) of sorbent with 100 mL of water containing sodium fluoride at 10 mg/L as initial fluoride concentration.The mixture was agitated in a thermostatic shaker at a speed of 250 rpm at room temperature.The defluoridation studies were conducted for the optimization of various experimental conditions like contact time, initial fluoride concentration, adsorbent dose, and fluoride uptake with fixed dosage by varying pH.All of the experiments were carried out at 30˚C ± 5˚C temperature.Fluoride ion concentration was measured with an automatic ion analyzer using the equipment model Metrohm-TitriC based on ion chromatography principle and the pH of the solution was also measured by Orion ion selective equipment.All other water quality parameters were analyzed by using standard methods [66].The effects of different dosages of biosorbent on fluoride uptake were studied by keeping the initial fluoride concentration and volume of the solution steady at 10 mg/L and 100 mL, respectively, at pH 4.
The fluoride concentration is retained in the adsorbent phase, qe (mg/g), was calculated according to [45], ( ) where qe is the amount of fluoride adsorbed (mg/g), Co and Ce are the original/initial and residual concentration at equilibrium (mg/L), respectively, of fluoride in solution, w is the weight of bi-adsorbents and V is the volume of solution.

Theory of Isotherm Models
The abilities of two widely used isotherms such as the theoretical Langmuir and empirical Freundlich isotherms, to model the adsorption equilibrium data were examined.These models allow for an explanation about the nature of adsorption from the shape of the curve.Langmuir adsorption isotherm [45] is perhaps the best known of all isotherms, which is often applied in solid/liquid systems to describe the saturated monolayer adsorption.The linear form of the equation is used to describe the adsorption capacity for a particular range of adsorbate concentration.It can be represented as: where Ce is the equilibrium concentration (mg/L), qe is the amount of ion adsorbed (mg/g), qm is a constant for a complete monolayer (mg/g), and Ka is an adsorption equilibrium constant (L/mg).The constants qm and Ka can be determined from a linearized form of Equation ( 2) by the slope of the linear plot of Ce/qe versus Ce.
The Freundlich adsorption isotherm [67] is based on adsorption on a heterogeneous surface and is the earliest known relationship describing the adsorption equilibrium.The linear form of this equation is given by: where qe is the amount of ion adsorbed (mg/g), Ce is the equilibrium concentration (mg/L), K F and 1/n are empirical constants indicating the adsorption capacity and adsorption intensity, respectively.The plot of log qe versus log Ce of Equation (3) results in a straight line and from the slope and intercept of the plot, the values for n and K F can be obtained.

Instrumental Analysis
The surface morphology and size distribution of the micro carbon fiber and nanocomposites were analyzed by means of scanning electron microscopy (SEM) and transmission electron microscopy (TEM).The crystallinity and structure of the nanocomposites and the as-prepared activated carbon from grass were analyzed by X-ray diffraction (XRD).

XRD, SEM, and TEM Analysis for Nanocomposites
The X-ray diffraction patterns have demonstrated that there were no peaks to indicate a crystalline structure (Figure 1(a)) of thermally treated grass at 600˚C.The absence of diffraction peaks confirms this material consists of amorphous fibers.XRD patterns of iron oxide and titania induced fiber showed a number of diffraction peaks for each which confirma crystalline product.In the case of iron oxide, the XRD shows the presence of magnetite, iron nitride and iron carbide (Figure 1 maze of interconnecting channels, quite large at the surface and terminating in small channels in the interior [57].This is indicative of carbon that is a good adsorbent base for the adsorption of pollutants from wastewaters.From the above studies, we observed that these metal oxides have able to convert micro carbon fiber into nanocarbon fiber, and the same process oxides have been changed into nanoparticles of different shapes.Scanning electron microscopy (SEM) and TEM images of the iron nanocomposite showing small nanoparticles were distributed on the surface of the carbon fiber homogenously, decorated with nanoparticles of size 5 to 20 nm in diameter.The small size and uniformity of these nanoparticles was maintained after the 800˚C treatment; apparently the strong interaction between the cabonfiber and the nanoparticles prevented their aggregation.This process is very simple and scalable in both the case.The embedded metal nanoparticles on the surface of cabon from TEM images are clearly affirming that a solid state thermal chemical reaction has been occurred between carbon fiber and metal oxides during the process.There was no much distance between the metal nanoparticles and the carbon surface which is indicating that the carbon has taken place in the chemical reaction.It is possible that the carbon fiber did chemically participate in the reaction by reducing the activation energy for the transformation of the metal oxides.This will require more detailed investigations in future studies.Nevertheless the proximity of metal nanoparticle to the carbon surface is expected to alter the surface charge of the carbon fibers, which can be probed via zeta potential study.Results of micro carbon fiber, IBNC and TBNC have been explained as under adsorption section.SEM of the fluoride treated IBNC surface is shown in Figure 3(d), which is different from the untreated fluoride surface and has confirmed that fluorides are adsorbed on the surface regardless of surface irregularities.

Effect of Contact Time and Initial Fluoride Concentration
Contact time plays a very important role in the adsorption dynamics.The effect of contact time on adsorption of fluoride onto fiber is shown in Figure 4. Batch adsorption studies using the concentrations 2.0, 3.0, 4.0, 6.0, 8.0 and 10.0 mg/L of fluoride solution and with 0.5 g of the adsorbent were carried out at 30˚C ± 5˚C as a function

Effect of pH and Adsorbent Dose
The effect of pH on removal of fluoride was examined (Figure 6).Test mixtures containing initial fluoride ion concentration of 10 mg/L and 0.5 g/100mL of adsorbents were adjusted to various pH values both in acidic and alkaline ranges, mixed for 60 min and analyzed for residual fluoride ion concentration by ion chromatography.Observations from Figure 6 were made at pH levels of 2, 4, 6, 8, and 10 and these experiments were conducted separately for all three adsorbents.The pH was maintained at the desired value within ±0.2 by adding 0.  HNO 3 or 0.1 N NaOH.The percentage of fluoride ions removed decreases with increase in pH for all three adsorbents.At a pH of 2 to 4, which gives the maximum fluoride removal for both cases of metal oxide based fiber, about 97% and 92% of fluoride ions are removed for IBNC and TBNC, respectively (Figure 6).
The effect of adsorbent dosage on the removal of fluoride ions was examined by varying the adsorbent concentration from 0.1 g/100mL to 0.6 g/100mLwhile all other parameters were kept constant i.e., the initial fluoride ion concentration was 10 mg/L and the pH was 4. The percentage of removed fluoride ion increases with an increase in adsorbent concentration for all three adsorbents, but after sometime, it gradually approaches a saturation value, denoting an equilibrium as shown in Figure 7. IBNC removes 87% of fluoride at a dose of 0.1 g/100mLand increases to 97% removal at 0.5 g/100mL (Figure 7).The removal of fluoride ions increases from 84% to 92% when increasing the concentration of TBNC from 0.1 g/100mL to 0.5 g/100mL at a temperature of 35˚C ± 5˚C (Figure 7).

Adsorption Isotherms
The equilibrium data isotherm analysis for fluoride adsorption onto the processed fiber from grass at pH 4.0 ± 0.2 and 35˚C ± 5˚C are shown in Figure 8. Results indicate that the adsorbent has a high affinity for fluoride ion adsorption under these conditions.As the concentration increases, adsorption capacity of fluoride also increases, but a plateau was gradually reached for all three materials when the concentration of fluoride is beyond 7 mg/L.The equilibrium data has been analyzed by linear regression of isotherm model equations, Freundlich and Langmuir (Figure 9    values of the slopes and intercepts of the respective linear plots [69].The present data fit the Freundlich models (Figure 9) well (R 2 > 0.99).The average adsorption capacity (LogK F ) obtained for IBNC, TBNC, and fibrous carbon (from grass) is 0.41, 0.72, and 0.71 respectively, while the adsorption intensity (n) is found to be 2.39, 1.39, and 1.38, respectively.It is observed from the comparison of adsorption isotherms between IBNC, TBNC, as well as fiberous carbon (from grass) that both adsorption capacity and adsorption intensity for IBNC is found to be higher.With n > 1, there is support for the favorable adsorption of fluoride onto these biosorbents.The linear plot for the Langmuir isotherm is shown in Figure 10.The relationship of linearity represented by Ce/qe versus Ce shows a correlation coefficient, R 2 , of 0.97 which is partially deviating from linearity.The present data is well fit to the Freundlich isotherm inferring multilayer adsorption is taking place on the surface of the biosorbent, due to ionic interactions.This finding is further explained by using adsorption potential theory [70] which explains that adsorptive forces originating from the well-known Van der Waals interaction can attract molecules at distances greater than a single molecular diameter.This force helps to reinforce the cohesive interaction between adsorbing molecules.Therefore, adsorbate molecules tend to concentrate at high energy sites on the modified fibrous carbon surface (nano carbon fiber) and undergo precipitation.The adsorption potential is highest in cracks and pores of the adsorbent, thus explaining the extensive adsorptive powers of fibrous carbon with its highly developed pore structure.Figure 11 represents a heterogeneous carbon surface and the broken lines represent the equipotential surface of having different energies (E), with the highest adsorbate concentration close to the surface, decreasing as one progresses away into the bulk of the solution.This is analogous to the atmospheric pressure gradient in a gravitational field.At pH 4, zeta potential of the adsorbent surface for IBNC (ξ = +4.07mV), TBNC (ξ = −0.72 mV), and micro fibrous carbon (ξ = −5.56mV) is found to be less negative as shown in Figure 12.Therefore, the electrostatic interaction of fluoride ion is different depending on the altered surface charge.IBNC has a high adsorption capacity and high adsorption intensity that may be due to dominating electrostatic interactions as well as Van der Wall forces existing between adsorbate and adsorbent.
IBNC cultivates a highly porous structure within each granule of carbon.The wall of these pores provides internal surface area.The surface area ranges from 500 to 1500 m 2 /g resulting in the extreme sorptive properties of these materials.Although adsorption is favorable in all cases, the adsorption capacity is following the sequence where IBNC > TBNC > micro fiberous carbon, depending on the proximity to the surface.The more cracked and porous structure with positive surface potential has a high affinity to fluoride uptake as compared to others.The Freundlich model, which is an indicator of surface heterogeneity of the adsorbent, explains the observed phenomena with the experimental findings [71] [72].

Conclusion
The defluoridation studies of the biosorbent, i.e. processed Bermuda grass (Cynodon dactylon), have been carried out in batch mode.The most excellent defluoridation occurred at the optimum time of 60 minutes allowing for a success rate as high as 96% while keeping 10.0 mg/L fluoride concentration, 0.5 g dosage of adsorbent,  and pH 4 constant.Thus, these materials show superior adsorptive efficiency as compared to previously studied defluoridation using natural adsorbents [8]- [10].The presence of iron oxide particles is effective for the removal of fluoride using this adsorbent.The sorption of fluoride using this adsorbent followed the Freundlich model and the sorption process was found to be spontaneous in nature.Based on the above said description, iron oxide nanoparticle-induced biosorbent could be used to remove fluoride selectively from water.

Figure 1 .Figure 2 .
Figure 1.The XRD of the micro carbon from grass (a); The iron oxide based nanocomposite (b); The XRD of titania based nanocomposite (c).

Figure 2 (
b) showed that the particles of iron oxide grafted to carbon giving rise to a highly irregular surface with large microspores and Figure 2(c) showed that the titania grafted carbon has irregular surfaces with various microstructures, including rods and spheres.TEM shown in Figure 3(a) and Figure 3(b) presents agglomeration of different particle sizes with spherical shapes showing both macro and micropores, and a few voids and crevices.In contrast, the TEM of titania grafted carbon, in Figure 3(c), showed a multi-modal distribution of particles size and solid titania rods are distributed irregularly [68].Figure 3(c) also clearly reveals that the outer and inner diameter of the TiO 2 rods is around 8 -10 and 5 -7 nm, respectively, and they have an open-end structure.

Figure 4 . 4 .
Figure 4. Effect of contact time on removal of fluoride on 0.5 g of adsorbents in water. of time to evaluate the defluoridation and adsorption rate constants.The adsorption of fluoride increases with time and gradually attains equilibrium after 60 minutes as shown in Figure 4. Therefore 60 minutes was fixed as minimum contact time for the maximum defluoridation of the sorbent.The adsorption of fluoride decreased from 99.5% to 92% by increasing fluoride concentration from 2.0 to 10.0 mg/L as shown in Figure 5. Further, it was observed that the removal curves are continuously decreasing indicating the possibility of the formation of multilayers of fluoride ion at the pore volume and interface of the adsorbent.

Figure 5 .
Figure 5.Effect of initial fluoride conc.on removal of fluoride on 0.5 g of adsorbents in water.

Figure 6 .
Figure 6.Effect of pH on removal of fluoride for 0.5 g of adsorbents in water.

FluorideFigure 7 .
Figure 7. Effect of adsorbents dosage on removal of fluoride at Ph: 4 and contact time 60 min in water.

Figure 8 .
Figure 8.Effect of concentration on fluoride removal.

Figure 9 .Figure 10 .
Figure 9.Effect of concentration on fluoride removal fits to Freundlich model.

Figure 12 .
Figure 12.Effect pH on surface potential of adsorbents.