Spectrophotometric determination of fluoride in drinking water using aluminium complexes of triphenylmethane dyes

A sensitive spectrophotometric determination of fluoride in drinking water has been developed using aluminium complexes of triphenylmethane dyes (chrome azurol B and malachite green) as spectrophotometric reagents. The method allowed a reliable determination of fluoride in the range of 0.5–4.0 mg∙l-1 for chrome azurol B and 0.0–2.0 mg∙l-1 for malachite green. The molar absorptivity for the complexes of chrome azurol B at 582 nm and malachite green at 622 nm is 1.44 × 104 and 2.56 × 104 l∙mol-1∙cm-1, respectively. The sensitivity, detection limit, quantitation limit, and percentage recovery for 1.5 mg∙l-1 fluoride for the method using chrome azurol B were found to be 0.125 ± 0.003 μg∙ml-1, 0.2 mg∙l-1, 0.5 mg∙l-1, and 97.1 ± 4.2, respectively, and for malachite green were 0.143 ± 0.002 μg∙ml-1, 0.1 mg∙l-1, 0.3 mg∙l-1, and 97.9 ± 4.1, respectively.


Introduction
Fluoride (F -) occurs in almost all waters from trace to high concentration (Dar et al., 2011).It has been shown to cause significant effects in humans through drinking water (WHO, 2006).Low concentrations of fluoride in drinking water have been considered beneficial to prevent dental carries (Maliyekkal et al., 2008;Quin et al., 2009), but excessive exposure to fluoride in drinking water can give rise to a number of adverse effects (WHO, 2006;Armienta and Segovia, 2008;Aldrees and Al-Manea, 2010;Arveti et al., 2011;Dunne and Verrel, 2011).WHO has set a limit value of 1.5 mg•ℓ -1 for fluoride in drinking water (WHO, 2004;Rafique et al., 2008).There is a narrow margin between the desired and harmful doses of fluoride in drinking water (Czarnowski et al., 1996;Jha et al., 2011).Therefore, an accurate, simple, rapid and cost-effective analytical method is of high importance.
Spectrophotometric methods are widely used in the determination of fluoride because of advantages such as simplicity, convenience, accuracy and reproducibility (Zolgharnein et al., 2009).They are based on the reaction of fluoride with coloured metal chelate complexes, producing either a mixed-ligand ternary complex or replacement of the ligand by fluoride to give a colourless metal-fluoride complex and the free ligand with a colour different to the metal-ligand complex (Einaga and Iwasaki, 1981).
Triphenylmethane (TPM) dyes of most value today are those with auxochromic and bathochromic groups like amino and hydroxyl in para position to the methane carbon (Witterholt, 1969).TPM dyes with groups capable of holding metals in stable combination, such as eriochrome cyanine R and azurol S, have been used successfully as spectrophotometric reagents for fluoride determination (Thrun, 1950;Macnulty et al., 1956;Sarma, 1964;Dixon, 1970;Einaga and Iwasaki, 1981;Kiernan, 1984).TPM dyes absorb strongly in the visible region to produce intense, brilliant shades of red, violet, blue, and green (Witterholt, 1969).They are characterised by high molecular extinction coefficients and by 2 bands in the visible absorption spectra.The longer wavelength (X band), which corresponds to an oscillation of the charge in the X direction, is of greater intensity than the shorter-wavelength (Y band), which corresponds to an oscillation of the charge in the Y direction (Witterholt, 1969;Zollinger, 2003).
The present study aimed to develop spectrophotometric methods for determination of fluoride in the range of 0.0-2.0mg•ℓ -1 , compatible with the WHO limit value of 1.5 mg•ℓ -1 , using aluminium complexes of TMP dyes such as chrome azurol B and malachite green as fluoride spectrophotometric reagents, due to their spectrophotometric properties described above.

Experimental Instrumentation
Beckman DU-7500 single beam spectrophotometer with 1.0 cm quartz cells was used for wavelength scanning and for spectral studies.Hitachi U-1500 UV/Vis single beam spectrophotometer with 1.0 cm quartz cells was used for the absorbance measurements at fixed wavelength.

Reagents
Chrome azurol B provided by Fluka, malachite green carbinol base provided by Aldrich, and aluminium chloride hexahydrate

544
provided by Fluka were used without any further purification.All the chemicals were of analytical reagent grade except where stated otherwise.Solutions were prepared using doubledistilled water.Chrome azurol B and malachite green ligand solutions and their aluminium complex solutions were prepared using ethanol from Merck (96%).Standard fluoride stock solution was prepared by dissolving 0.1382 g of sodium fluoride provided by Merck in 250 mℓ water.The stock solution was further diluted as needed.

Preparation of the metal complexes solutions
Job's method of continuous variation was adopted for the determination of the composition of the coloured complex (Werner and Boltz, 1971a;Werner and Boltz, 1971b).Aluminium to ligand (chrome azurol B or malachite green) ratio was also studied by preparing complexes with the most common molar ratios (1:1, 1:2, 1:3, 2:1, 3:1, 2:3, and 3:2) to enable comparison between the spectra of these different complexes.The blank was prepared by the same procedure using the solvent instead of the aluminium ionic solution.
Aluminium to ligand ratio was found to be 1:2 for both chrome azurol B and malachite green.Thus, the complex solutions for the spectrophotometric measurements were prepared as 1:2 ratios from aluminium and ligand of 1 × 10 -4 M in ethanol solutions, and the solutions were then diluted to the concentration (≈5 × 10 -5 M) suitable for the spectrophotometric measurements.

Reaction of fluoride with the prepared complex solutions
Various amounts of fluoride were added in the range 0-2 mg•ℓ -1 to a 25 mℓ volumetric flask containing aluminium complex solution of chrome azurol B or malachite green in ethanol (5×10 -5 M, 24.5 mℓ).The solution was made up to volume with water.The absorbance was measured at the wavelengths of maximum difference (425, 581 nm for chrome azurol B; 428, 622 nm for malachite green) in the electronic spectra between the ligand and the complex.The spectra for the reaction of different amounts of fluoride with the complex were compared.

Determination of fluoride in real drinking water samples
The method under investigation was tested using 3 real drinking water samples which had been collected and analysed by the Central Public Health Laboratory of the Ministry of Health, Palestine, which is responsible for managing water quality.Samples were collected in June 2011 from 2 groundwater wells in Tulkarm District (Nazlet Issa well and Abu Sabha well in the village of Atteel) and 1 groundwater well in Tubas District (Aqaba well).Fluoride was analysed colourimetrically using SPADNS as fluoride reagent and a Hack -DR/2010 spectrophotometer.Nitrate, sulphate, chloride and other characteristic data of the 3 samples are given in Table 1.According to the Ministry of Health, fluoride ranged between 0.2 and 0.8 mg•ℓ -1 in the West Bank groundwater resources (Salem, 2011).Therefore, fluoride was measured after spiking the water samples with 0.5 mg•ℓ -1 fluoride.The results obtained were then compared with those reported by the Central Public Health Laboratory using the SPADNS method (Table 1 and Table 2).

Chrome azurol B and its aluminium complexes
Chrome azurol B is commonly known as 2 // , 6 // -dichloro-4 / -hydroxy-3-3 / -dimethyl fuchsone-5, 5 / -dicarboxylic acid; solochrome azurine B; and omega chrome azurine B (Gurr, 1971).It is used in the spectrophotometric determination of uranium, palladium, copper, etc. (Gregorowicz et al., 1983;Jancar et al., 1989;Guo, 1992), and makes stable complexes with different elements (Uesugi and Shigematsu, 1977;Boodts et al., 1982).Chrome azurol B is dark orange in ethanol.Figure 1 shows its electronic absorption spectra; it displays 1 band with 2 heads at 280 and 296 nm in the UV region, and a main broad band of greater intensity with heads at 444 and 580 nm in the visible region.The molar absorptivity at 444 and 580 nm is 13 830 ± 64 and 12 520 ± 68 ℓ• mol -1 •cm -1 , respectively.The main absorption band is due to a Π→Π* transition, while the electron donors generally cause strong bathocromic shift (Zollinger, 2003).In general, TPM dyes exhibit 2 bands in the visible absorption spectra (Witterholt, 1969).In the ligand under investigation, the 2 band transitions overlap to produce a single band with only a shoulder on the shorter wavelength side.The greater the fraction of the positive charge on the auxochromes is the longer the wavelength of the absorption.However, there is hypsochromic shift (decrease in the wavelength) in the Y band of chrome azurol B due to the ability of the auxochromes (carboxylic and hydroxyl group) to eliminate the charge by resonance.
Aluminium chrome azurol B ratio was determined as 1:2.The complex exhibits a dark pink colour in ethanol and displays 1 band in the visible region with 2 heads at 538 and 582 nm (Fig. 1).The molar absorptivity at these 2 heads is (1.40 ± 0.023) ͯ 10 4 and (1.44 ± 0.026) ͯ 10 4 ℓ•mol -1 •cm -1 respectively.There is a bathochromic shift of about 94 nm after complexation with aluminium.The stability of the complex in ethanol solution was examined for 1 month, as shown in Fig. 2, and the complex is stable.A possible chemical structure of aluminium chrome azurol B 1:2 complex is given in Fig. 3.The sensitivity of the colour reaction between metals and chrome azurol B is enhanced by the presence of surfactants (Uesugi and Shigematsu, 1977).N-cetyl-N,N,N-trimethylammonium bromide was examined as a surfactant to enhance the colour of the investigated complex.However, the results showed that it has no significant effect at the spectrum of the complex.

Malachite green and it aluminium complex
Malachite green is a cationic basic dye which has found a widespread use as colorant in industry and as microbial agent (Green, 1990;Eldem and Özer, 2004;Rajgopal et al., 2006).It is also used as a spectrophotometric reagent for determination of dissolved phosphate in both water and soil extracts (Motomizu et al., 1983;Linge and Oldham, 2001).Malachite green reaches equilibrium between its cationic and the colourless carbinol forms at pH 10.1, whereas the complete ionisation to the salt occurs at pH 4.0 or lower (Goldacre and Phillips, 1949;Golding et al., 1998).
Malachite green displays 2 bands in the visible region in ethanol, at 428 and 622 nm (Fig. 4).The long wavelength band provides a blue component to the colour, while the short band provides a yellow component to the colour.The combination of these two components of the colour is the colour of the dye, which appears green to human eyes.The long wavelength band is characterised by an oscillation of an electron cloud across the molecule between the two auxochromes, while the shorter band corresponds to an oscillation through the phenyl group (Mason and Nord, 1951;Green, 1990;Zollinger, 2003).
The results obtained from applying Job's method of continuous variation indicated the 1:2 aluminium complex.The  546 complex is possible because the lone electron pair of the nitrogen atom in the malachite green is delocalised into the outer orbitals of the Al 3+ ion, and a compound which is analogous to a donor-acceptor complex is formed (Minczewski et al., 1975).The proposed complex has a Π-bond, between the donor (nitrogen atom) and the acceptor (Al 3+ ion), which increased the binding energy of the central Al atom.In general, the electron transition between the donor and the acceptor was not complete (Minczewski et al., 1975).
Aluminium malachite green 1:2 complex is very dark green in ethanol.It displays 2 bands in the visible region at the same wavelength as the ligand, at 428 and 622 nm (Fig. 4), with a hyperchromic effect, which leads to an increase in the absorption after complexation with aluminium.The molar absorptivity at 622 nm is (2.56 ± 0.045) ͯ 10 4 ℓ•mol -1 •cm -1 .
The stability of aluminium malachite green 1:2 complex was examined over 1 month.Figure 2 shows that the complex is stable at 622 nm at high concentrations such as 6.64 x 10 -4 M. On the other hand, the complex is only stable for a few hours at a concentration below 1.2 ˟ 10 -5 M (Fig. 5) due to the hydrolysis of the complex and release of the free ligand.This results in a change in the colour of the ligand to light green.

Reaction of fluoride with aluminium chrome azurol B complex
Fluoride reacts with the dark pink aluminium chrome azurol B complex to produce a colourless aluminium fluoride complex by replacement of the chrome azurol B by fluoride and liberation of the free ligand.This leads to a change in colour from that of the complex, dark pink, to that of the free ligand, dark orange (Fig. 6), according to the equation below.Aluminium reacts with fluoride to give compounds of the nature of (AlF 6 -3 ) or (AlF y (OH) 6-y ) -3 (Macnulty et al., 1956).
The absorption spectra of the reaction of fluoride with the Al chrome azurol B complex showed that fluoride interacts to cause an increase in absorbance of the aluminium complex at 425 nm and a decrease in absorbance at 581 nm, due to the formation of the aluminium fluoride complex and release of the free ligand.Figure 7 shows that the absorbance of the aluminium complex is related linearly at 425 and 581 nm to the concentration of fluoride in the range 0.0 to 4.0 mg•ℓ -1 ; the squared correlation coefficient, R 2 , is 0.9904 and 0.9907, respectively.The equation of the linear calibration curve at 425 and 581 nm is y = 0.0885x + 0.1808 and y = −0.1251x+ 0.8861, respectively.
The sensitivity, detection limit, limit of quantification, and recovery, of 1.0, 1.5, and 2.0 mg•ℓ -1 F of the Al chrome azurol B complex for the spectrophotometric determination of fluoride at 581 nm, are given in Table 2.The sensitivity was taken as the average of the slope of the calibration curve for 5 replicates.The detection limit and the limit of quantification were calculated as 3.3σ/S and 10σ/S, respectively, where σ is the standard deviation of response and S is the slope of the calibration curve.The recovery was measured as the average of 10 replicates.Fluoride was measured using the proposed field method in 3 real water samples, and, because of the low fluoride content in the West Bank's groundwater, the samples were spiked by 0.5 mg•ℓ -1 fluoride.The recovery of fluoride by the proposed field method is given in Table 2, and is in agreement with that reported by the Central Public Health Laboratory using SPADNS as fluoride colorimetric reagent.
The interference studies were done by measuring the influence of anions such as chloride, nitrate, and sulphate on the determination of 1.0 and 1.5 mg•ℓ -1 fluoride.Therefore, the expected interfering anions were added in such concentrations commonly found in the natural water (Table 3).Nitrate and chloride were added in the range of 0-100 and 0-500 mg ℓ -1 , respectively.The data on interference in Table 3 show that nitrate concentrations up to 20 mg•ℓ -1 do not interfere with the determination of fluoride.However, higher results for fluoride are obtained when the amount of nitrate is 20 mg•ℓ -1 or more.Chloride at concentrations up to 100 mg•ℓ -1 does not interfere (Table 3).Large amount of chloride (>100 mg•ℓ -1 ) may be overcome by adding an excess of silver perchlorate to the solution (Hensley and Barney, 1960).
Sulphates interfere with most visual and photometric methods for determination of fluoride.Sulphate interferes through competition with fluoride to form a complex with the metal, thereby resulting in higher concentrations (Price and Walker, 1952;Ruzicka et al., 1966).In the present work, when the amount of sulphate is higher than 100 mg•ℓ -1 sulphate interferes with the determination of fluoride by increasing the absorption at 425 nm and decreasing the absorption at 581 nm.This can be overcome by precipitating sulphate in cold solutions by the addition of aqueous barium chloride solution and aqueous agar-agar solution, and then separating the precipitate by filtration (Dixon, 1970).
Interference may also be due to the presence of metal ions that give a colour with the dye or to the presence of cations, e.g., iron, zirconium, magnesium, which form complexes with fluoride in competition with aluminium (Nishimoto et al., 2001;Dixon, 1970).However, metal ions from scarce sources are not expected in drinking water.The stability constants of the metal fluoride complexes increase in the following order: and Reilly, 1963).When the sample solution is expected to contain a large amount of these cations, and because of the high tendency of quinalizarin to form colour chelates with various metal ions (Snell, 1978;Srivastava and Banerji, 1967), it is recommended that the metal interferences are masked by adding complexing agents such as DCTA (trans-1,2-aminocyclohexane-N,N,N´,N´tetraacetic acid) (Nishimoto et al., 2001) or EDTA (ethylenediaminetetraacetic acid (Snell, 1978).

Reaction of fluoride with aluminium malachite green complex
Fluoride interacts with aluminium malachite green 1:2 complex to cause a decrease in the absorbance of the aluminium complex at 622 nm.The absorbance of the aluminium complex is related linearly at 622 nm to the concentration of fluoride in the range of 0.0-2.0mg•ℓ -1 (Fig. 7); the squared correlation coefficient, R 2 , is 0.9878 and the equation of the linear calibration curve is: y = −0.1427x+ 0.8372.The sensitivity, detection limit, limit of quantification, and recovery of 1.0, 1.5, and 2.0 mg•ℓ -1 F of the Al malachite green complex for the spectrophotometric determination of fluoride at 622 nm, are given in Table 2.The only interference comes from sulphate at a concentration of more than 250 mg•ℓ -1 , resulting in a positive error of about 20%.This error can be overcome by precipitation of sulphate using barium chloride (Dixon, 1970).

Conclusions
Two members of the triphenylmethane dye group were examined as spectrophotometric reagents for fluoride determination.Promising results were obtained with aluminium 1:2 complexes of chrome azurol B and malachite green.These two complexes can be used as spectrophotometric reagents for fluoride in the ranges 0.5-4.0 and 0.3-2.0mg•ℓ -1 , respectively.Aluminium chrome azurol B complex can be recommended as a sensitive reagent but with low selectivity, while aluminium malachite green complex can be recommended as a fluoride reagent at concentrations higher than 1.5 x 10 -5 M.