Anal. Chem. 2001, 73, 5590-5595
Speciation of Aluminum in Rainwater Using a Fluoride Ion-Selective Electrode and Ion-Exchange Chromatography with Fluorometric Detection of the Aluminum-Lumogallion Complex Hirokazu Hara,* Hidemasa Kobayashi, Mayumi Maeda, Ai Ueno, and Yohko Kobayashi
Department of Chemistry, Faculty of Education, Shiga University, Otsu, Shiga 520-0862, Japan
Soluble aluminum in rainwater was separated into three categories: free aluminum (Al3+), fluoride complexes (sum of AlF2+ and AlF2+), and other forms of aluminum. The free form of the aluminum ion (Al3+) was directly obtained from the separation data of aluminum species according to their charge using gradient elution cationexchange chromatography. The aluminum fluoride complexes were estimated by combining the data of the free and total fluoride determined using a fluoride ion-selective electrode, with the assumption that 2+ charged aluminum species consisted only of AlF2+. The rest of the aluminum species had a 1+, neutral, or negative charge and mainly consisted of organic complexes. The origin of the organically bound aluminum is discussed. The concentration range of the total dissolved fluoride and aluminum in the rainwater samples was usually in the micromolar to submicromolar range, and the ratio of [T-F]/[T-Al] was found to be between 1 and 4. The speciation of dissolved aluminum into three categories was carried out on the basis of data of 15 rainwater samples collected in the city of Otsu. The speciation of aluminum in natural water becomes an important research area in many fields, especially in environmental chemistry.1 The chemical and biochemical nature of aluminum, that is, toxicity, dissolution and transport mechanism, bioavailability, etc., is dependent on its chemical form. The free and fluoride complex of aluminum showed a higher toxicity than the citrate complex for certain kinds of fish in acidic (pH 4.4) water.2 The residual aluminum in drinking water is reported to enhance the risk of Alzheimer’s disease.3 Many studies on the speciation of aluminum have already been published.4 Driscoll proposed an analytical procedure for the fractionation of aqueous aluminum into three categories: acid* Fax: +081-077-537-7840. E-mail:
[email protected]. (1) The Environmental Chemistry of Aluminum, 2nd ed.; Sposito, G., Ed.; CRC Press: Boca Raton, FL, 1996. (2) Driscoll, C. T.; Baker, J. P., Jr.; Bisogni, J. J.; Schofield, C. L. Nature 1980, 284, 161-164. (3) McLachlan, D. R. C.; Bergeron, C.; Smith, J. E.; Boomer, D.; Rifat, S. L. Neurology 1996, 46, 401-405. (4) Pyrzynj ska, K.; Bulska, E.; Guc¸ er, S.; Hulanicki, A. Chem. Anal. (Warsaw) 1999, 44, 1-14.
5590 Analytical Chemistry, Vol. 73, No. 22, November 15, 2001
soluble, nonlabile monomeric, and labile monomeric.5 His procedure has influenced the speciation studies of aluminum in natural waters.6 However, the procedure provides only operational results. The speciation method in which individual chemical species are directly determined may be more desirable. 27Al NMR was used to detect polynuclear aluminum species in forested Spodosol soil horizon samples.7 The direct application of 27Al NMR to the speciation of aluminum in natural waters is believed to be rather difficult because of its low sensitivity. On the other hand, fluorometric detection has a high sensitivity for the analysis of aluminum8 in the submicromolar concentration range. The method was combined with cation-exchange chromatography for the separation of aluminum species according to their charge.9,10 Sutheimer and Cabaniss used lumogallion as a fluorescent reagent and applied the method to lake water analysis.11,12 They ascribed the second peak (2+ charged species) to AlF2+ and possibly other Al organics, but no direct evidence was provided, because their interest was mainly focused on the first peak (1+, neutral, or negatively charged species). To understand the effect of the acidification of rainwater, the speciation of aluminum in rainwater is very important. The sensitivity of HPLC with fluorometric detection is sufficient for rainwater analysis. Recently, we proposed a speciation method of fluoride in rainwater using a fluoride ion-selective electrode.13 The sensitivity was extended down to 1 × 10-8 M (1 M ) 1 mol dm-3) using an acid buffer.14 Fluoride in rainwater was determined as total and free fluoride, which corresponded to the determined values obtained with and without an aluminum-chelating reagent in a buffer solution. The difference between the total and free fluoride should be regarded as the sum of 2 × [AlF2+] and [AlF2+]. Thus, the result may be applied to the speciation of aluminum. (5) Driscoll, C. T. Int. J. Environ. Anal. Chem. 1984, 16, 267-283. (6) Fairman, B.; Sanz-Medel, A.; Gallego, M.; Quintela, M. J.; Jones, P.; Benson, R. Anal. Chim. Acta 1994, 286, 401-409. (7) Hunter, D.; Ross, D. Science 1991, 251, 1056-1058. (8) Sutheimer, S. H.; Cabaniss, S. E. Anal. Chim. Acta 1995, 302, 112-121. (9) Jones, P. Int. J. Environ. Anal. Chem. 1991, 44, 1-10. (10) Tsunoda, K.; Yagasaki, T.; Aizawa, S.; Akaiwa, H.; Satake, K. Anal. Sci. 1997, 13, 757-764. (11) Sutheimer, S. H.; Cabaniss, S. E. Anal. Chem. 1995, 67, 2342-2349. (12) Sutheimer, S. H.; Cabaniss, S. E. Geochim. Cosmochim. Acta 1997, 61, 1-9. (13) Hara, H.; Yabuuchi, K.; Higashida, M.; Ogawa, M. Anal. Chim. Acta 1998, 364, 117-123. (14) Hara, H.; Huang, C. Anal. Chim. Acta 1997, 338, 141-147. 10.1021/ac010428w CCC: $20.00
© 2001 American Chemical Society Published on Web 10/06/2001
Figure 1. Schematic diagram of ion-exchange chromatographic system.
In this paper, we propose a new method for the speciation of dissolved aluminum in rainwater by combining both methods using cation-exchange HPLC and the fluoride ion-selective electrode. The suggested method can be applied to the speciation of dissolved aluminum in rainwater into three categories: free, fluoride, and other complexes. The relationship between the separated peaks of aluminum and the actual chemical species is discussed. EXPERIMENTAL SECTION Reagents. All chemicals were guaranteed reagent grade and were obtained from Nacalai Tesque (Kyoto, Japan) unless otherwise specified. Nitric acid was for the harmful metal analysis. Calcium nitrate was purchased from Sigma (St. Louis, MO). Oncedistilled water was further purified using a Millipore Milli-Q Labo system. A 2.2 M β-alanine/HCl buffer with and without 1.1 × 10-2 M CDTA (1,2-cyclohexanediamine tetraacetic acid) was used for the determination of the total and free fluoride. The postcolumn reagent solution was prepared by dissolving 0.05 mM lumogallion (DOJINDO, Kumamoto, Japan) with warming in 0.2 M acetate buffer (pH 5.2). Two eluents for the gradient elution were used: pH 4.0 HNO3 and 0.0165 M Ca(NO3)2 (the pH was also adjusted to 4 by HNO3). Aluminum standards were prepared in plastic volumetric flasks. A stock standard (10-2 M) was prepared by dissolving 0.4744 g of AlK(SO4)2‚12H2O into 100 mL of 10-3 M nitric acid. The pHs of the 0-10 × 10-7 M aluminum standard solutions were adjusted to ca. 4.0 by adding 0.1 mL of 0.1 M nitric acid into 100 mL volumetric flasks. These standard solutions were stable for at least one month in the plastic volumetric flasks. Aluminum fluoride solutions were prepared by mixing a suitable amount of 1 × 10-4 M aluminum and fluoride standard solutions. The pH was also kept around 4.0 by adding 0.1 M nitric acid. Apparatus. Figure 1 shows a schematic diagram of the HPLC system. The basic scheme was the same as that of Sutheimer and Cabaniss.11 Two Shimadzu LC-10AD chromatographic pumps were used to create a concentration gradient of the calcium nitrate eluent solution. The sapphire plunger of each pump was replaced by a zirconia ceramic plunger in order to decrease the possibility of aluminum elution from the pump. The sum of the flow rates
was kept at 0.9 mL min-1. A single plunger pump (Nihonseimitsu model SP-T-2501-U, Tokyo, Japan) was used to deliver the reagent solution. The pulse was effectively damped using a pulse damper (GL Science model HPD-2, Tokyo, Japan). The flow rate was set to 0.4 mL min-1. To remove residual aluminum ions in the eluents, a precolumn was placed just after each main pump. The precolumn was prepared by filling a chelating resin into a PEEK column (4.7 mm i.d., 15 cm long). A Chelex 100 (Bio-Rad) or TSK-8HQ was used as the chelating resin. The latter was based on a TSK-Gel TOYOPEARL HW-65F (Catalogue No. 07465, Tosoh, Tokyo, Japan), on which 8-hydroxyquinoline was covalently fixed. We prepared the TKS-8HQ by the method reported by Landing et al.15 We preferred the TSK-8HQ column over the Chelex 100 column because of its excellent efficiency in removing aluminum ions and its long-term stability. The rotary injection valve was made of stainless steel (Rheodyne model 7010, Cotati, CA) fitted with a PEEK 500-µL injection loop. For the separation of aluminum species according to its charge, an ion-exchange column (Mitsubishi Kagaku MCI-Gel, SCK01, 6-mm i.d., 50 mm long) was used. This column is made of a copolymer based on polystyrene having a sulfonic acid functional group. Two PEEK mixing coils (0.5-mm i.d., 5 m long) were used, the first maintained at 60 °C in a water bath, while the second remained at room temperature. PEEK tubing was used in the whole system with some exceptions (otherwise, a poly(tetrafluoroethylene)(dPTFE) tube was used). Fluorescence was detected by a Shimadzu RF-10Axl detector equipped with a 150 W xenon lamp, a 12 µL quartz flow cell, and a photomultiplier tube. The excitation and emission wavelengths were 500 and 590 nm, respectively. A chromatogram was recorded using a Shimadzu chromatopac CR6A. The calculation of the peak area was performed using Shimadzu COALA CH-10 chromatographic data processing software. An Orion fluoride ion-selective electrode (model 94-09BN) was used in combination with a double-junction reference electrode (model 90-02).13,14 The 1:10 dilution of the buffer solution without (15) Landing, W. M.; Haraldsson, C.; Paxe´us N. Anal. Chem. 1986, 58, 30313035.
Analytical Chemistry, Vol. 73, No. 22, November 15, 2001
5591
CDTA was used as an outer filling solution. An Orion ionalyzer model EA920 was used for the measurement of the electrode potentials. Determination of Free and Total Fluoride. The potentiometric determination of fluoride was carried out by the conventional batch method. A 25-mL portion of sample solution was placed into two PTFE beakers, and 2.5 mL of the respective buffers with and without CDTA was added. The electrode potentials of these solutions were measured first in order to select four suitable concentrations from the series of fluoride standard solutions. The determination was then made from the calibration graph, which can be regarded to be linear at least down to 3 × 10-7 M. If the calibration graph was curved in the range below 3 × 10-7 M, the third-order Spline function was used for the interpolation.16 Determination of Total Dissolved Aluminum. Total dissolved aluminum was determined by the same HPLC system without the separation column. Samples were injected through a 0.22-µm disk filter (Millipore Millex-GS). A calibration graph was prepared using 0, 2, 4, 6, 8, and 10 × 10-7 M aluminum standard solutions. If the concentration exceeded 1 × 10-6 M, 0, 5, 10, 15, and 20 × 10-7 M standard solutions were used. The correlation coefficient of the calibration graph usually exceeded 0.999. The blank aluminum level calculated from the intercept of the calibration graph was usually below 4 × 10-8 M. For the calculation of a sample concentration, this blank level was ignored. However, the accuracy of the determination on the order of 10-8 M may be relatively poor, because it is quite difficult to reduce the blank level any more. Separation of Aluminum Species On the Basis of Its Charge. The flow rate of two eluents, that is, pH 4.0 HNO3 and 0.0165 M Ca(NO3)2 (pH 4.0) was adjusted to 0.082 and 0.818 mL min-1, respectively. After mixing, the concentration of calcium ions was 0.015 M. One minute before the sample injection, the flow rates of both pumps were exchanged. Then 1.5 min after the sample injection, the gradient elution begins. During 2 min, the flow rates were linearly changed to the preset values with time. Namely, the flow rate of the pH 4.0 HNO3 changed from 0.082 mL min-1 to 0.818 mL min-1 instantaneously, remained for 2.5 min, then back to 0.082 mL min-1 after 2 min and vice versa. This gradient elution procedure produces three distinctive peaks for an aluminum fluoride solution, the same as those in ref 11. Those peaks were assigned to species having a charge of 1+ or less, 2+, and 3+. The peak assignment was confirmed to agree with the theoretical calculation of species distribution.11 The final 3+ peak contains not only Al3+ but also aluminum hyroxides. The middle 2+ peak may contain a small fraction of free Al(OH)2+; however, the contribution of this fraction in the 2+ peak was ignored in this study for the sake of simplicity. The first 1+ peak represents species with a negative, neutral or 1+ charge; however, it will be denoted as the 1+ peak. The baseline of a separated chromatogram was recalculated so that the total peak area became as close as that of total dissolved aluminum. Rainwater Sampling. Rainwater was sampled in a polyethylene bottle using a 30-cm-diameter plastic funnel on the roof of the Natural Science building, Faculty of Education, Shiga University. Sampling was carried out during every rainfall, but all of the
rainwater was also collected for intermittent rains. The collected rainwater was filtered using a 0.45-µm membrane filter (Millipore HAWP04700) using a plastic filter holder and was stored in a refrigerator. The sampled rainwater was analyzed within 1 week. In a preliminary experiment, the determined values for aluminum and fluoride were almost unchanged in some samples during the week. The sampling period was from March to December, 2000. Equilibrium Calculations. The equilibrium distribution of aluminum species and free fluoride was calculated using an original program named AlFx coded in BASIC language. In this program, the percentage distribution of the aluminum species and free fluoride concentration can be calculated from the pH, total aluminum ,and total fluoride concentrations. The equilibrium constants for aluminum hydroxide and aluminum fluoride were taken from Table 1 of ref 11. The equilibrium constants of aluminum fluorides were newly determined by the authors.11 The aluminum hydroxide species were hypothesized to appear in the 3+ peak, as was done in ref 11. The equilibrium calculation for aluminum fluorides using the program AlFx were confirmed to agree with the theoretical calculation in Figure 3(a) of ref 11, which was made using Titrator.17 To examine the effect of an organic ligand, a program named AlFxOx was prepared in which the equilibrium calculation of two
(16) Hara, H.; Wakizaka, Y.; Okazaki, S. Talanta 1987, 34, 921-926.
(17) Cabaniss, S. E. Environ. Sci. Technol. 1987, 21, 209-210.
5592 Analytical Chemistry, Vol. 73, No. 22, November 15, 2001
Table 1. Result of Five Independent Determinations of Aluminum Fluoride Solutiona
no.
[T-Al] 10-7 M
1 2 3 4 5 mean SD RSD (%)
10.00 9.59 10.01 10.18 10.12 9.98 0.23 2.3
species distribution, % 1+b 2+ 3+
[F-F] 10-6 M
5.0 5.7 4.9 4.9 5.0 5.1 0.34 6.6
0.416 0.403 0.337 0.428 0.340 0.385 0.043 11.2
54.0 55.8 56.7 56.0 57.4 56.0 1.3 2.3
41.0 38.4 38.5 39.1 37.6 38.9 1.3 3.3
a Al:F ) 1:1, [T-F] ) 10-6 M. b The neutral and negatively charged species (if any) are contained here.
Figure 2. Equilibrium distribution of aluminum species with variation in the concentration ratio of fluoride to aluminum ([T-F]/[T-Al]). Alf ) Al3+ + Al (OH)x(3-x). Dashed lines represent species distributions calculated with the formation constants reported in ref 11. [T-Al] ) 10-6 M, [T-F] ) 0.125, 0.25, 0.5, 1, 2, 4, and 8 × 10-6 M. pH 4.0.
Figure 3. Equilibrium concentration of free and total fluoride with variation in the concentration ratio of fluoride to aluminum ([T-F]/[TAl]). Experimental conditions are the same as those in Figure 2. Dashed lines represent the results of the theoretical calculations with the formation constants reported in ref 11.
ligands system, that is, fluoride and oxalate, was carried out. This program is a simple extension of the program AlFx. The association constants between aluminum ion and oxalic acid were taken from ref 18. RESULTS AND DISCUSSION Determination of Fluoride and Aluminum. The accuracy of the fluoride recovery in aluminum fluoride solutions (Al:F ) 1:1, 2:1, 5:1, and 10:1; [F-] ) 10-6 M) was first examined. The determined values of fluoride were 1.017, 1.079, 1.025, and 1.013 × 10-6 M, respectively. The result shows that the total fluoride can be accurately obtained even for a large excess of aluminum ion relative to fluoride. Table 1 gives the precision of the percentages of the 1+, 2+, and 3+ charged species of aluminum and free fluoride for five independent measurements (day-to-day variation). The precision for aluminum was satisfactory as compared to that for free fluoride. The reproducibility for both the free and total fluorides depends on the concentration. The RSD (%) for three successive measurements was usually less than 3% in the micromolar range, but it may increase above 10% below 3 × 10-7 M. The RSD (%) was usually better for the total fluoride than for the free fluoride determinations. The electrode potential slowly decreased in the measurement of free fluoride of some rainwater samples. This result may be due to the slow change in equilibrium between aluminum and fluoride caused by the addition of the buffer solution. In the determination of free fluoride, this effect was taken into account. The precision of the aluminum percentage for the five repeated measurements (variation within a day) was also determined. The RSD (%) for 1+, 2+, and 3+ peaks of aluminum fluoride solution (Al:F ) 1:1, [F-] ) 10-6 M) was 5.4, 2.2, and 3.7%. Analysis of Aluminum Fluoride Solutions. Three distinct peaks can be observed from the chromatographic separation of the aluminum fluoride solutions. The 1+ peak consists of AlF2+ and AlF3; and the 2+ peak, of AlF2+. To examine if the separation was carried out quantitatively, the result was compared with the theoretical expectations. (18) Jaber, M.; Bertin, F.; Thomas-David, G. Can. J. Chem. 1977, 55, 36893699.
Figure 2 shows the equilibrium distribution of aluminum species with variation in the concentration ratio of fluoride to aluminum ([T-F]/[T-Al]). Free aluminum Alf consists not only of Al3+ but also the sum of aluminum hydroxides (Al(OH)2+ + Al(OH)2+ + Al(OH)3 + Al(OH)4-). The determined percentages can be said to fit the theoretical expectations on the whole. The small discrepancy between the experimental and theoretical values in the case of the 1+ and 2+ peaks was also reported in ref 11. This discrepancy may be due to the inaccuracy of the set of equilibrium constants for AlFx, especially in the concentration region of [T-F]/ [T-Al] > 2. Figure 3 shows the equilibrium distribution of the free (and total) fluoride with variation in the concentration ratio of fluoride to aluminum([T-F]/[T-Al]). The determined percentages of the free (and total) fluoride were in fair agreement with the theoretical calculations. The speciation method using a fluoride ion-selective electrode confirmed a correct value for the complexed fluoride as the difference between the total and free fluoride, (T - F)F. Speciation Procedure. For the speciation of rainwater using the fluoride and aluminum data, the calculation scheme is proposed as follows: First, the following equation is valid if the ratio [T-F]/[T-Al] is not very large, that is, if [AlF3] and [AlF4-] are negligible.
(T - F)F ) [AlF2+] + 2[AlF2+]
(1)
Because the ratio [T-F]/[T-Al] was usually
