A study on the defluoridation in water by using natural soil

Abstract

Removal of excess fluoride (F⁻) from the water has been attempted by several authors by using different materials both natural and artificial. The main aim of this paper was to attempt the fluoride removal by using the locally available red soil adopting column method. The red soil was mixed in different proportion with sand in order to increase the porosity and permeability property of the medium. It was optimized for 4:1 ratio of red soil to sand and it was used for the following experiment. The experiment was conducted in 11 batches for a period of about 9,213 min. Fresh standard solution of F was used in each batch, prepared from Orion 1,000 ppm solution. The samples were collected and analyzed for pH, EC (Electrical Conductivity) and HCO₃. Rate of flow of water and efficiency of adsorption were calculated and compared with the fluoride removal capacities of the medium. The medium used for the fluoride removal was subjected to FTIR analysis before and after the experiment. The variation of IR spectrum before and after treatment signifies the changes in the OH bonding between Al and Fe ions present in the soil. The variation in pH decreased during the course of defluoridation. Higher F removal was noted when flow rate was lesser. An attempt on the regeneration of the fluoride adsorbed soil was also made and found to be effective.

Introduction

Excess of fluoride in groundwater has become a threat in recent days due to the lesser availability of potable groundwater resource. There are different methods of defluoridation proposed by different authors, like ion exchange, precipitation and adsorption. Among these methods, adsorption is a widely used method for defluoridation of water because of its easy method of operation and cost effectiveness. Some of the adsorption materials broadly used for defluoridation are; physico-chemically treated sand (Togarepi et al. 2012), microwave assisted activated carbon (Dutta et al. 2012), aluminum sulfate treatment (Malay and Salim 2011), pumice (Malakootian et al. 2011) and raw Bauxite (Sajidu et al. 2012). An extensive survey on the removal of excess fluoride in water shows that different techniques has been attempted by several authors, using natural and synthetic material (Table 1). This review of the literature shows that the study of fluoride removal by natural materials/soil is less. One such attempt was made (Chidambaram 2000; Chidambaram et al. 2003) for removal of fluoride in groundwater by using natural materials and showed that, untreated charcoal seems (Table 2) to have lesser effect on the concentrations of fluoride with time and it adsorbs only 0.5–1 mgl⁻¹ of fluoride. The brick and fly-ash show typically a sudden removal of fluoride to 6.6 and 5.6 mgl⁻¹ respectively. >50 % reduction of concentration takes place within 30 min. Red soil absorbs 9 mgl⁻¹ of fluoride immediately after 15 min. After 30 min the fluoride decreases from 0.09 mgl⁻¹ and keeps on reducing with time up to 0.035 mgl⁻¹ in 90 min to 0.039 mgl⁻¹ at 120 min. In serpentine the fluoride removal was found to be maximum from 0 to 15 min, i.e. 3.7 mgl⁻¹: later the defluoridation capacity decreases to 5.8 mgl⁻¹ at 120 min. But the defluoridation capacity is less when compared to red soil. The study reveals that among these five materials used for defluoridation red soil has a good defluoridation capacity followed by brick, serpentine, fly-ash and charcoal. Near equilibrium is attained after about 30 min of the experiment in red soil and brick (Table 2). Maximum defluoridation occurs immediately after the experiment has started in almost all the materials used except for lesser impact on charcoal. The molality of fluoride removed per gram is higher in red soil when compared to the other materials. Red lateritic soil has the highest fluoride removal capacity because it has oxide of aluminium and iron as its major components. In general aluminum compounds are found to be good fluoride removers because of reaction between Al⁻ and F⁻ molecules (Chidambaram et al. 2003). Hence, both Al⁻ and Fe⁻ in red soil are found to be a good removers of fluoride. The column study thus reveals that the quantity of materials used were according to their effective fluoride removal capacity. But this study lacks detailed monitoring of data over a periodic time interval, where sequential changes can occur rapidly; further, the materials were not examined/studied before and after the treatment. Keeping these factors in mind, red soil which is an effective fluoride remover was used in this study to understand the fluoride removal capacity over time. They are also studied for their change in their bonding nature before and after the treatment. Further an attempt was also made to understand their regeneration property/capacity after this experiment.

Table 1 The removal of excess fluoride in water has been attempted by several authors

Table 2 Fluoride in time intervals (all the values are in mgl⁻¹)

Materials and methods

A column was set up for the study by using mixture of red soil and sand with a ratio by weight of 4:1. It is noted that the red soil is fine grained substance with clay size particles and, with good ion exchange capacity but lesser permeability, hence it was mixed with sand to increase the permeability. The removal rate of fluoride was tested using 250 gm of mixture soil (200 g of red soil and 50 g of sand). The fluoride concentration in the samples was measured by using an Orion, model 2100 expandable ion analyzer EA 940 and the fluoride ion selective electrode BN 9609 (Orion, Thermo Fisher Scientific, USA). The standard solution of 10 mg/l fluoride was prepared from 100 mgl⁻¹ standard (Thermo Orion) by using DDW (Doubled Distilled Water). The samples were collected at definite intervals. Eleven different Batch experiments were conducted by passing Fluoride standard of 10 mgl⁻¹. 10 ml of sample was collected after the treatment and analyzed. A total of about 62 samples were collected in different batches during the entire experimental period. Out of these samples collected, different parameters were observed along with F concentration. Rate of flow of water through the column, their variation in pH, Percentage of fluoride adsorption, adsorption capacity, bicarbonate concentration and EC were also observed. pH and EC were analyzed by ion sensitive electrodes, HCO₃ was estimated by titrimetry method using standard procedures (APHA 1998).

The adsorption capacity was calculated by using the formula,

$$q_{\text{e}}(\text{mmol}/\text{g})=(\text{Co}-\text{Ce})\times V/W\times 1,000$$

Ce is equilibrium adsorbate (fluoride) concentration (mmol/l). Co is initial adsorbate (fluoride) concentration (mmol/l). q_e amount of fluoride adsorbed per unit gram of the adsorbant at equilibrium (mmol/g). V is volume of solution (l). W is weight of adsorbant (g).

Results and discussion

FTIR

Fourier Transform Infrared (FTIR) spectroscopy is an established experimental technique for determining qualitative mineral identification and is currently being developed for quantitative mineralogy. The majority of the natural earth materials are silicates, carbonates, phosphates, etc. The elementary building unit of the silicate minerals is the 4 SiO tetrahedron having a net–4 charge. The interlinking of the 4 SiO tetrahedra results in formation of various polymers (Makreski et al. 2005). The structural classification of the silicate minerals is in fact based on the extent of sharing of oxygen anions between the adjacent tetrahedra. It makes the silicates the largest and the most complicated class of minerals. An important associate ion of the silicate mineral group is OH.

Fe–Al and OH bondings

The intensity and especially the frequency of the IR band in the ν (OH) region could serve as a tool for reliable estimation of the Fe and/or Al content. Fe and Al are the dominant ions in the red soil; the strength of O–H bond slightly differs when different cations are coordinated to the hydroxyl groups as it is depicted in the FTIR spectrum (Table 3). A simple harmonic model of different M³⁺–O oscillators (M = Fe and/or Al in red soil) may result in frequency shifting of the band in the spectrum (Makreski et al. 2007). The effect of Fe substituting for Al at the non-OH coordinated M1 and especially M3 octahedral site causes an increase in the bond strength at the O₄ atom ([M2] Al–O₁₀–H···O₄–[M1] Al [M1] Al [M3] Fe) for Fe it is principally due to the difference in the electronegativity with respect to Al (Della Ventura et al. 1996). Since the maximum of the OH stretching vibration in our IR spectrum is observed at 3,450 cm⁻¹. ν(OH) vibrations in the OH bearing minerals are generally observed in the 3,600–3,400 cm⁻¹ region, the band at 3,627 cm⁻¹ probably arise from FeOH stretching, whereas the strongest band at 3,450 cm⁻¹ has OH and perhaps somewhat 2HO stretching mode character.

Table 3 Assignment of bands observed in FTIR spectrum

Sorosilicate in FTIR spectra

The presence of the bands in many sorosilicate spectra studied (Denning et al. 1972; Lazarev 1972) at frequencies 1,170–1,080 cm⁻¹ much above those from nesosilicates (Makreski et al. 2005) implies that these bands could be assigned as ν as (Si–O_b–Si) modes (O_b denotes bridging oxygen). Therefore, the origin of the band at 1,088 cm⁻¹ is not questionable and should be attributed to the mentioned Si–O vibration. According to Wang et al. (1994) the symmetric stretching vibration of Si–O_b–Si type should be expected in the 750–450 cm⁻¹ region, this mode was prescribed to the strong band at 695 cm⁻¹.

The assignment of the bands in the 700–500 cm⁻¹ region was much more complicated because, in addition to the Si–O–Si deformation modes, this region is also typical for the OH vibrational modes. It causes the bands in this region to be overlapped and even partly coupled because the OH modes were expected to appear in wide region. Taking these considerations into account, the weak bands at 795, 695 and 586 cm⁻¹ were assigned as predominantly bending Si–O–Si vibrations whereas the strong band beyond 586 cm⁻¹ (absent in the studied non-hydroxide pyroxenes) probably is a result of out-of-plane bending OH vibrations.

Variation in the FTIR Spectra of Muscovite, Phlogophite, Actinite and Aragonite minerals were identified by FTIR. The Phlogophite observed very strong band at 465.31 cm⁻¹ (Si–O–Si) vibration. The Quartz observed strong band at 463.57 cm^−1. The Muscovite observed very strong band at 1038.25 cm⁻¹ in the Si–O vibration. The Montmorillonite is observed at 1,629 cm⁻¹ in the H–O–H bending vibration. The Kaolinite is observed Lattice band at 3627.40 cm⁻¹and Al–O–H stretching at 3,620–3,692 cm⁻¹. The maxima at 3,704 cm⁻¹ and 3,627 cm⁻¹ represent the OH stretching with the following cation combinations Mg²⁺Mg²⁺Mg²⁺, Mg²⁺Mg²⁺Fe²⁺, Mg²⁺Fe²⁺Fe²⁺ and Fe²⁺Fe²⁺Fe²⁺. The substantial broadening of the series of bonds by substitution of Fe³⁺ by Al³⁺ in other amphiboles on, the smaller six-fold coordinated Fe³⁺ cations which gets generally concentrated in M2 sites studied (Wilkens 1970; Hawthorne 1983; Skogby and Annersten 1985). Therefore, the lowest frequency OH stretching peaks at 3,523 and 3,450 cm⁻¹ are prescribed to the Fe²⁺Fe²⁺Fe³⁺ and Fe²⁺Fe³⁺Fe³⁺ cation combinations, respectively, being in accordance with the literature data for Actinolite mineral (Mustard 1992). The ν (OH) region has shown that studied mineral contains both Mg²⁺ and Fe²⁺ cations.

Sediments which are Mg²⁺ rich and poor in Fe²⁺ show absorptions at 620 and 586 cm⁻¹ that are not found in others specimens. According to Stubican and Roy (1961a, b) the strong absorption is observed between 620 and 692 cm⁻¹ assigned to Si–O vibration the one near 550 cm⁻¹ to Si–O–Al ^VI vibration, the one near 760 cm⁻¹ to Si–O vibration and another near 830 cm⁻¹ to Si–O–Al vibration. It is also reported that an absorption due to H–O–Fe would appear near 812 cm⁻¹ which is absent in the spectra before and after treatment and one represented by Si–O–Fe ^VI vibration would be found near 495 cm⁻¹ (Stubican and Roy 1961b).

It is due to the fact that the pyroxene structure is built by the complex chains of 4 SiO anions whereas the 4 SiO building blocks in the nesosilicates are isolated. Such structural difference perturbs the degeneracy of the ν2, ν3 and ν4 modes in the pyroxene infrared spectra (Sterns 1974) causing the appearance of larger number of IR bands as a result of the different bonding between the terminal (Ot) and bridging (Ob) oxygen to the Si atoms. In this context, the highest wave number peaks were assigned as the Si–Ot rather than Si–Ob modes. It is due to their greater force constants compared to the Si–Ob modes, because the Si–Ob motions are additionally shared between the adjacent tetrahedra. On the other hand, the wave numbers of the bands in this region are nearly constant suggesting that chain vibrations are not particularly sensitive to the population of the cation sites by Rutstein and White (1971). It means that the observed bands in the 1,100-850 cm⁻¹ regions are, to great extent, due to pure ν (O–Si–O) and ν (Si–O–Si) vibrations.

For experimental quantitative and to understand changes in the adsorbent (if any) due to fluoride sorption, FTIR analysis were done before and after adsorption. As shown in Fig. 1, the FTIR spectrum of the samples presents significant spectroscopic change due to fluoride sorption. The broad band corresponding to 3,450 cm⁻¹ (range of 3,698–3,380 cm⁻¹) represents OH stretching vibrations that at 1,391 cm⁻¹ to Al H stretching and 1,013 cm⁻¹ represents the characteristic stretching bands of Si–O–Al. The band at 586 cm⁻¹ may be ascribed to the stretching of Al OH. On closer examination, it has been observed that the intensity of many of the peaks shows variations after fluoride sorption. Chukin and Malevich (1977) demonstrated that the treatment of SiO₂ sample with fluoride results in a decrease in the intensity of the (OH)⁻ band at 3,704 cm⁻¹ or its complete disappearance due to fluoride uptake. This is readily explained, since it is known that (OH)⁻ and F⁻ ions have closely similar dimensions and can isomorphously replace each other. So, to ascertain whether such an exchange reaction is taking place in the fluoride sorption onto absorbant, the ratio of peak heights of the unbounded surface (OH)⁻ band at 3,386 cm⁻¹ to that at 3,700 cm⁻¹ are compared before and after adsorption. Before adsorption, in the virgin adsorbent this peak height was 2.56, but after fluoride sorption, it is found to be 6 in the fluoride adsorbent. This shows that the OH⁻ band at 3,704 cm⁻¹ is decreasing due to fluoride sorption, confirming the exchange of OH ions, enhancing fluoride removal. In the similar way, the peak height of (OH)⁻ band at 3,627 cm⁻¹ to that of Al OH band at 586 cm⁻¹ is also compared to infer the changes before and after treatment.

Fig. 1

FTIR spectrum of the mixture before and after the experiment

Effect of pH in fluoride removal

To understand the fluoride sorption behavior under different pH values sorption was monitored, under (Kamble et al. 2007; Sarkar et al. 2006) the neutral, protonated and deprotonated sites. Behavior of the adsorbent towards fluoride removal mainly depends on the initial solution pH. The effect of pH on fluoride removal performance of the adsorbent was studied over a wide pH range of 3–11 by different authors (Togarepi et al. 2012; Dutta et al. 2012; Malakootian et al. 2011). For the same system, protonation and subsequent F⁻ sorption via ligand exchange may also explain F⁻ removal at pH <6.5, in accordance with following equations:

$$\text{M}-\text{OH}+{\text{H}}^{+}\to \text{M}-{\text{OH}}_2^{+}$$

(R1)

$$\text{M}-{\text{OH}}_2^{+}+{\text{F}}^{-}\to {\text{M}}^{+}-{\text{F}}^{-}+{\text{H}}_2\text{O}$$

(R2)

Where, M represents metal ions like Al³⁺, Fe³⁺. The supported iron and aluminum oxides form the aqua complex with water and form the surface charges through amphoteric dissociation. At acidic pH, more positively charged surface sites developed which attract the negatively charged fluoride ions by electrostatic attraction resulting in the enhanced fluoride removal at acidic pH (Denning et al. 1972; Lazarev 1972; Wang et al. 1994). Reactions (R1) and (R2) hold true for fluoride adsorption at acidic medium.

$$\text{M}-\text{OH}+{\text{F}}^{-}\to \text{M}-\text{F}+{\text{OH}}^{-}$$

(R3)

Where, M represents metal ions like Al³⁺, Fe³⁺. Reaction (R3) represents the ligand-exchange interaction between fluoride and hydroxyl ions at neutral pH. It is known that at pH <7, free bases are feebly ionized and reaction (3) does not proceed rapidly to the right as compared to the reactions (1) and (2) (Ramos et al. 1999). The pH increase of initial solution in the experiment is near to 10 which have no significance influence on fluoride removal efficiency. Further increase in pH above 10 drastically reduces the fluoride removal efficiency by 52 %, which may be due to the competition between hydroxide and fluoride ions for active sites in this pH range. The decrease in adsorption at higher pH value (pH>10) may also be possible due to abundance of (OH)⁻ ions which leads to increased hindrance to diffusion of fluoride ions by Stubican and Roy (1961b). Exchangeable Ca²⁺ present in soil at near neutral pH has been reported by some workers (Mustard 1992; Stubican and Roy 1961a). In all the cases, decreased F⁻ removal at high pH may be due to the existence of an hydroxyl envelope on solid particles causing negative surface charge, which would effectively repel F⁻ from the sorption sites and simultaneously compete with them (Rutstein and White1971). At the later stage of the experiment the pH of the treated water shows increase of (H⁺) concentration by decrease in pH values M-OH₂ + F⁻ → M-F⁻ + H⁺ + (OH)⁻. At higher pH value, a slight decrease in fluoride removal was observed. Similar observation in the reduction of fluoride removal capacity at alkaline pH ranges may be attributed to the competition from the negatively charged (OH)⁻ ions by Ghorai and Pant (2005). A sharp decrease in fluoride removal was noted (Denning et al. 1972) and they attributed it due to the formation of the weakly ionized HF (pKa = 3.2) at low pH values and due to the competitiveness of the (OH)⁻ and F⁻ ions in the bulk, at high pH values.

EC and HCO₃ variations

The variation in the batches indicates that the electrical conductivity generally decreases after the batch reactions. It shows increasing trend in batch 8 and 11. This may be due to the regeneration process during the flow of water and saturation of the available exchange sites. Though bicarbonate concentration shows fluctuation but generally exhibit a decreasing trend except in batch 8, where both the HCO₃ and EC shows increasing trend.

The study shows that rate of flow ranges maximum from 0.02 to 0.83 mls⁻¹ with an average of 0.22 mls⁻¹. The pH shows maximum from 9.50 to minimum 6.63 with an average of 7.69, the percent of adsorption shows maximum from 96 to minimum BDL (Below Detection Limit) with an average of 86.45 (Table 4). Adsorption capacity ranges with a maximum of 0.0038 and a minimum BDL with an average of 0.0025, the HCO₃ ranges with a maximum from 707.60 mgl⁻¹ to a minimum of 24.40 mgl⁻¹ with an average of 182.80 mgl⁻¹ and the EC ranges with maximum from 4354 μs/cm to a minimum of 291 μs/cm with an average of 723.52 μs/cm. The correlation analyses of the parameters were carried out and these parameters show that the rate of flow of water has positive correlation to pH and negative to HCO₃, pH has negative correlation to all other parameters, with strong negative correlation to fluoride adsorption capacity and HCO₃ (Table 5). This indicates that the efficiency of fluoride removal increases when HCO₃ concentration decreases. The efficiency of fluoride removal is higher when the adsorption capacity is more. The adsorption capacity shows a poor positive correlation to electrical conductivity of the solution.

Table 4 The maximum, minimum and average values of rate of flow, pH, adsorption, HCO₃ and EC

Table 5 Correlation analysis for the parameters observed during the experiment

Removal efficiency

Different batch experiments were conducted based on the mixture appropriation earlier carried out. For the same set of mixture F⁻ Standard of 10 mgl⁻¹was prepared for each batch. 5 ml of sample was collected and analyzed. The average concentration of almost all the batches were around 0.5 (i.e.) the fluoride removal efficiency was 95 % but the efficiency of removal starts decreasing at batch 11 and 12. It is evident that after 5,946 min i.e. after 99 h and its further reduced to 90 % after 146 h. Hence for a known volume of 946 cm³ of materials the efficiency decreases from 99 h and as time progresses the pH become more acidic. In general, it was noted that the initial treated water collected from each batch had high pH and F⁻ concentration which tends to reduce during the reaction.

Fluoride removal capacity was determined by the Weber–Morris equation, the removal capacity varies during the experimental period. The rate of flow (q) ranges from 0.04 to 0.36, but the initial readings of each batch may even reach zero. The decrease of q starts from 4,489 min, until then the value almost remains constant. Plot shows the plot of rate of flow (q) vs. t^1/2 (Fig. 2). It was observed that the plots are linear over the whole time period and the plots exhibit a plateau, indicating that the external surface adsorption is dominant at the initial stage and stage of intra-particle diffusion is attained and continued up to 4,489 min (half of the experimental period) after which the equilibrium was attained. The anions are slowly transported via intra-particle diffusion into the pores of the adsorbent and finally retained in the particles. Also it was observed that the initial linear portion of the curve does not pass through the origin and later stage of fluoride adsorption does not obey Weber–Morris equation, which indicates that mechanism of fluoride adsorption is rather complex process and the intra-particle diffusion is not the only rate-limiting step (Thakrea et al. 2010). In general the removal capacity reduces from the initial time period its also interesting to note that at the beginning of each batch the removal capacity is lesser. This may be due to high rate of flow and lesser residence time for the fluoride to get adsorbed on to the material.

Fig. 2

Relation between rate of flow (mls⁻¹) and fluoride removal

Rate of flow

It is observed that at a lower flow rate fluoride removal was efficient, at least in the initial stage of process, which may be due to the adequate contact time between adsorbent and adsorbate in solution. However, the extent of fluoride removal decreases with gradual occupancy of active sites of adsorbent. The rate of flow of the water through the medium is higher during the initial time periods with no definite trend and it is lesser in the last few batches. Figure 2 shows that there is a considerable decrease of pH and increase of HCO₃ concentration at the end of the experiment. At higher flow rate, fluoride removal capacity decreases and breakthrough becomes steeper and fluoride removal capacity is lesser due to shorter residence time of solute in the column, because of which, fluoride solution elutes the column before fluoride adsorption equilibrium occurs.

Regeneration studies

The reusability of the adsorbent was studied in order to check the residual fluoride uptake capacity. The reusability experiment was carried out by repeating the adsorption experiment on used and dried sample mixture under similar condition. The sediments after the experiment were washed with distilled water then dried in atmospheric conditions. Later the above experimental set up was repeated by the regenerated sediments.

The regenerated red soil to sand 4:1 was taken in the column and the experiment was repeated by passing 10 mgl⁻¹of fluoride standard solution. Initial 10 mgl⁻¹ of F⁻ was passed through the column the F⁻ value reduced from 10 to 2.5 mgl⁻¹ with a rate of removal of 0.007 at 314 min and further the rate of removal decreased to 0.003 at 1,410 min. In the 2:2 the rate of removal was 0.072 till 127 min and it got decreased to 0.028 and 0.035 at 296 and 328 min, respectively. It is interesting to note that the F⁻ concentration decreases at varying rate and increases after 21 h of the experiment in the regenerated soil.

Conclusions

The treatment of fluoride rich waters by using natural materials shows that there is a significant variation in FTIR spectrum before and after treatment. The variation is noted in the OH region of the spectrum, indicating adsorption and variation in bonding strength. The sites with Fe-OH and Al–OH bonds present in the red soil plays a main role in the controlling the efficiency of fluoride removal. Fluoride removal is effective in near neutral conditions. The competitiveness of the OH⁻ and F⁻ ions can be attributed to the change in pH of the solution. The effective removal of fluoride is maintained for a longer period and decreases with time after the occupation of the active sites. Though the rate of flow is higher at the initial stages the removal was effective due to the availability of more active sites. Subsequently the rate of flow was lesser and still the removal was found to be effective due to the increase of the contact time between the adsorbent and the liquid. The regeneration of the medium after the experiment for about 9,213 min was attempted and found to be effective, this also helps in the field for daily backwash of the column after the treatment and help to regenerate itself. The results obtained indicate this as a possible method for the removal of Fluoride for an effective rural water supply scheme. Further due to its control over the rate of water flow, column experiments is proved to be an easy method for transferring the technology to the field. This method requires no power supply since water moves down by the gravitational force hence it becomes more cost effective and easy to handle by local community as a suitable green and clean technology for rural drinking water supply in Fluoride affected areas.