- Email: [email protected]
Long-term monitoring of arsenic and selenium species in contaminated groundwaters by HPLC and HG-AAS
The Science of the Total Environment 258 Ž2000. 171᎐181
Long-term monitoring of arsenic and selenium species in contaminated groundwaters by HPLC and HG-AAS Michael Raessler U , Bernhard Michalke, Sigurd Schulte-Hostede, Antonius Kettrup GSF National Research Centre for En¨ ironment and Health, Institute of Eco-logical Chemistry, P.O. Box 1129, D-85758, Neuherberg, Germany Received 15 January 2000; accepted 12 May 2000
Abstract The long-term concentration and distribution of species of arsenic and selenium in contaminated groundwaters from Kelheim was monitored. Most of the groundwater wells contained elevated concentrations of iron, manganese and sulfur. Arsenic ŽIII., arsenic ŽV., selenium ŽIV. and selenium ŽVI. were separated using high performance liquid chromatography ŽHPLC. based on phosphate buffers and collected in fractions. Due to the complex matrix, the fractions were analyzed element-specifically by hydride-generating atomic absorption spectrometry ŽHG-AAS.. The combination of HPLC and HG-AAS was selected due to the authors’ intention of developing an easy-to-handle, but nonetheless reliable, method suitable for the long-term monitoring of species distribution in an almost routine way, and taking account of the threshold values of 10 grl for each element, indicated by German drinking water regulations. To enhance the reliability of the method, analytical quality control experiments were carried out. When applied to groundwater wells from Kelheim ŽGermany. they revealed that arsenic ŽV. and selenium ŽVI. were the dominating species. The presence of arsenic ŽIII. and selenium ŽIV. was assumed to be supported by organic matter. 䊚 2000 Elsevier Science B.V. All rights reserved. Keywords: Arsenic; Selenium; Speciation; Long-term monitoring; Quality control
U
Corresponding author. Max-Planck-Institute for Biogeochemistry, Tatzendpromenade 1A, D-07745 Jena, Germany. Tel.: q49-3641-64-3812; fax:q 49-3641-64-3710. E-mail address: [email protected] ŽM. Raessler.. 0048-9697r00r$ - see front matter 䊚 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 8 - 9 6 9 7 Ž 0 0 . 0 0 5 3 5 - 0
172
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
1. Introduction Arsenic and selenium may have toxic effects on biota in groundwater samples. Humans may be affected if the concentrations are high. For this reason, the German drinking water regulation ŽVerordnung ueber Trinkwasser, 1996. indicates a threshold value of 10 grl for each element. Arsenic is supposed to induce cancer ŽSmith et al., 1992., and elevated concentrations of selenium may lead to selenosis ŽValentine et al., 1994.. However, it is known today that the toxicity of an element is determined by the chemical form in which it is present in the sample. For example, organic arsenobetaine, the most common arsenic compound in fish, is non-toxic to humans, whereas arsenic ŽIII. is more toxic than arsenic ŽV.. In case of selenium, selenium ŽIV. is considered more dangerous to aquatic organisms than selenium ŽVI. due to its higher solubility and bioavailabilty. Consequently, for adequate risk assessment, modern environmental research must be no longer restricted to the determination of the total concentration of an element, but must account for the different ‘species’ of this element present in the sample. It is the aim of speciation to reveal the identity and the quantity of different specific compounds or oxidation states of an element. For speciation to be quantitative, the sum of the concentration of the element in all species present must be equal to the total concentration of the element in the sample. Unfortunately, in most cases, speciation is still restricted to a relatively small number of very specialized research laboratories. Its application to the monitoring of groundwater, for example, is still far from being routine work. It was the authors’ intention to develop a simple, rugged and, nonetheless, reliable method for the speciation of inorganic arsenic and selenium and to apply it to groundwater wells from Kelheim, Germany. The samples were characterized by high concentrations of iron, manganese and sulfur due to leaching processes of pyrites and their oxidation products, iron oxides. Pyrites were formerly used for the production of sulfuric acid. Both pyrites and iron oxides were deposited in the Kelheim region or were
used as filling materials in road construction and railway embankments. Among other trace elements, pyrites and iron oxides contained elevated concentrations of leachable arsenic and selenium that penetrated into the groundwater. High concentrations of interfering ions in these samples made the direct determination of arsenic and selenium by ion chromatography based on conductivity or UV detection impossible. For this reason, a combination of HPLC and elementspecific detection was mandatory. HG-AAS was selected for the following reasons: Ž1. its high selectivity towards arsenic and selenium; Ž2. the fact that the determination of arsenic and selenium is almost free of matrix interferences since hydride-forming elements ŽAs, Se. are separated from the matrix prior to atomization; and Ž3. HG-AAS, which offers detection limits for arsenic and selenium of - 1 grl, which meet the needs for most environmental applications, is much less expensive than the more sophisticated ICP-MS. As the results of a reliability study and of analyses of Kelheim groundwaters show, the combination of HPLC and HG-AAS is well suited for the species-monitoring of arsenic and selenium in environmental samples.
2. Experimental section 2.1. Reagents The chemicals used throughout this procedure were of analytical grade or better. Deionized water was produced with Millipore deionizing cartridges. The mobile phase was prepared by dissolving 0.575 g NH4 H2 PO4 ŽAldrich., 1.32 g ŽNH4 .2 HPO4 ŽAldrich. and 2.13 g Na2 SO4 ŽFluka. in deionized water. The addition of Na2 SO4 was necessary for the quantitative recovery of selenate. This solution was transferred to a 1-l flask and filled to the mark. The pH of the eluent was adjusted to 5.9 by adding, dropwise, H3 PO4 85% ŽFluka.. Prior to use, the eluent was filtered through a 450-nm filter ŽMillipore. and degassed in an ultrasonic bath for 1 h. Stock solutions Žvolumes 100 ml, concentration s 1 grl, i.e. 1000 ppm. of arsenate, selenite
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
173
Table 1 Chromatographic system and operating parameters 1A Pump Injection valve Sample loop Column Žsize. Eluent Flow rate UV detector Wavelength Plotter Fraction collector
Beckman 114 M solvent delivery module Beckman 7725i ŽRheodyne. 100 l Hamilton PRP X-100 Ž250 = 4.6 mm. 5 mM ADP, 10 mM DAP, 15 mM Na2SO4, pH 5.9 1 mlrmin BioRad Model 1706 UVrVIS monitor s 195 nm Spectra Physics SP 4270 integrator BioRad Model 2110 fraction collector
1B Parameters for HG-AAS detection Model Hydride generator Cell temperature Excitation source Wavelength Purge I Purge II Reaction time Slit width
Perkin-Elmer 3030B MHS 20 1000⬚C EDL 193.7 nm ŽAs.; 196.0 nm ŽSe. 50 s ŽAs.; 20 s ŽSe. 40 s ŽAs.; 40 s ŽSe. 18 s ŽAs.; 12 s ŽSe. 0.7 nm
and selenate were prepared by dissolving 0.416 g Na2 HAsO4 ⭈ 7H2 O ŽFluka., 0.219 g Na2 SeO3 ŽAldrich. and 0.467 g Na2 SeO4 ⭈ 10H2 O ŽAldrich., respectively, in water. The stock solution of As ŽIII. was prepared by dissolving 0.132 g As2 O3 ŽAldrich. in 5 ml of 1 M NaOH and filling to the mark with water. Single ions and mixed ion wAs ŽIII., As ŽV., Se ŽIV. q Se ŽVI.x working standards were prepared prior to use by adequate dilution of the stock solutions. 2.2. Chromatographic separation The separation of species was carried out by HPLC. The chromatographic system and operating parameters are listed in Table 1a. Three column materials were tested for their ability of species separation: Nucleosil 100-5 NŽCH3 .2 ET 250r4 ŽMacherey and Nagel., Vydac 302 IC 4.6, 250 = 4.6 mm Žthe Separations Group. and Hamilton PRP X-100 ŽHamilton Corporation.. Whereas the stationary phase of the first two columns were silica-based, the stationary phase of the Hamilton column was made of a
polystyrol-divinylbenzene matrix which was stable in a broad pH range. Of the three columns tested, the Hamilton column showed the best separation efficiency under isocratic conditions and was, therefore, used for further experiments with model solutions and speciation analyses of groundwater samples. Retention times were as follows: arsenic ŽIII. 2.6 min; arsenic ŽV.; selenium ŽIV. 4.0 min; and selenium ŽVI. 11.1 min. The coelution of arsenic ŽV. and selenium ŽIV. was considered to be no drawback due to element-specific detection. For the quantitative recovery of selenate the ionic strength of the phosphate buffer eluent had to be enhanced by addition of 15 mM Na2 SO4 . Na2 SO4 was selected since it did not influence the pH of the eluent nor UV detection Žno signal at 195 nm.. UV detection was used for method development only. Without addition of sulfate, the recovery of selenate was - 80%. The overall separation lasted 15 min. Prior to quantification, the ions were collected in fractions. Five fractions were taken every 3 min. If present, As ŽIII. was found in fraction 1 Ž0᎐3
174
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
min., As ŽV. andror Se ŽIV. in fraction 2 Ž3᎐6 min. and Se ŽVI. in fraction 4 Ž9᎐12 min.. Fractions 3 Ž6᎐9 min. and 5 Ž12᎐15 min. did not contain either arsenic or selenium species. The collected fractions were immediately analyzed by HG-AAS.
3. Element-specific detection Instrumentation and operating parameters for the detection of species by HG-AAS are listed in Table 1b. A solution of 3% NaBH4 was used for hydride generation. The volume of each fraction was extended to 10 ml by adding 7 ml of 3% HCl ŽMerck.. Since all the arsenic or selenium in the sample was purged into the atomizer, addition of HCl was not a dilution of the sample. Standards and samples containing Se ŽVI. were reduced to Se ŽIV. prior to analysis by HG-AAS since Se ŽVI. did not form a hydride under the abovementioned conditions. A quantitative reduction was carried out by adding 5 ml of HCl 32% to 5 ml of the sample in Nalgene tubes. The tubes were sealed and exposed to a water bath at 80⬚C for 1 h. After reduction, the samples were cooled to room temperature and analyzed. The analyses were evaluated based upon calibration curves in the range 1.5᎐15 grl for each element. The detection limits of HG-AAS for arsenic and selenium were based upon the analyses of 10 blanks of HCl suprapur 3% ŽMerck.. The corresponding values were 0.6 grl ŽAs. and 0.5 grl ŽSe. with t Ž f s 9, Ps 95%, one-sided.. A 100-l sample loop was used for HPLC Žcf Table 1a.. The limit of determination of the overall method was given by the combination of both the detection limit of HG-AAS and the volume of the sample loop. Consequently, this was found to be 6 grl for arsenic and 5 grl for selenium. The total concentrations of arsenic and selenium were determined by both HG-AAS and ETV-ICP-MS to confirm the reliability of HGAAS. For the exact working conditions of ETVICP-MS, see Raessler et al. Ž1998.. A JY-70 ICP atomic emission spectrometer ŽJobin Yvon Instruments SA. was used to determine the Fe, Mn
and S concentrations in the groundwater samples. Analyses were carried out at s 259.9 nm ŽFe., 257.6 nm ŽMn. and 181.9 nm ŽS.. The concentration of SO42y was determined by ion chromatography ŽDionex DX 500.. The determination of DOC was carried out in the filtered samples using a high TOC analyzer ŽElementar. while the redox potential, pH and conductivity were determined with a WTW Multical instrument.
4. Quality control Model solutions containing single ions of As ŽIII., As ŽV., Se ŽIV. and Se ŽVI., as well as mixed standard solutions of all ions, were used to check the reliability of the chromatographic separation and the element-specific detection. Recoveries of the ions at concentration levels of 2 mgrl, 100 and 10 grl were determined and mass balances established for each ion. Besides the quantitative recovery, the stability of all species was evaluated by what we called ‘reinjection experiments’. Referring to arsenic ŽIII., reinjection was carried out as follows: the chromatographic separation and fractionation were carried out as described above. After the quantitative recovery of arsenic ŽIII. in fraction 1, an aliquot of this fraction was reinjected and the fractionation repeated. All five fractions were, once again, analyzed by HG-AAS. The reinjection experiment was successful, i.e. arsenic ŽIII. was assumed to be stable during the chromatographic separation if its recovery in fraction 1 was again quantitative and if there was no arsenic present in the other fractions. Analogous experiments were carried out for arsenic ŽV., selenium ŽIV. and selenium ŽVI..
5. Collection and preparation of groundwaters of the Kelheim region Groundwater samples were collected from seven different groundwater wells ŽA᎐G. at Kelheim, Lower Bavaria, Germany. The city of Kelheim is situated at the confluence of Altmuhl ¨ river and Danube river in the south-east of Germany. The sampling points were located in the
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
grounds of a chemical plant where sulfuric acid was produced; in the grounds of a cellulose-producing plant that had been shut down, and in the city center of Kelheim. Samples were taken during the winter, spring, summer and autumn seasons to allow monitoring of species throughout the year. Samples were pumped with a delivery of 1.2 lrs. To avoid contamination by the tubing, the pumping system was conditioned with well water. The first 50 l of each sample was discarded. Samples were filled into 1-l PE bottles, and stored at 4⬚C, protected from daylight. To avoid the formation and precipitation of iron and manganese oxyhydroxides, the samples were acidified to pH- 2 with HCl. The filled bottles were immediately sealed to avoid exposure to air. To avoid plugging of the HPLC valves, all samples were vacuum filtered through 0.45-m filtration membranes prior to HPLC separation. The separation and quantitation of species were carried out as described above.
6. Results and discussion 6.1. Quality control Quality control is of growing importance to speciation analyses ŽMichalke, 1999. which, in most cases, still suffer from a lack of adequate standard reference materials. The accuracy of results of speciation analyses is influenced by various parameters, such as sampling, sample storage and sample preparation. The separation of species and their analytical quantitation are the final steps of the speciation procedure. While it is almost impossible not to influence the distribution of species at least partly during the sampling procedure, every effort should be made to verify that there is no detrimental influence on the species that could be traced to separation and quantitation. This was the motivation for the reliability study results, which are presented here. More details referring to the quality control of this procedure are given by Raessler Ž1997.. Table 2a shows the results for the injection of single ions; and Table 2b the results for mixed ion
175
solutions at the 2 mgrl, 100 and 10 grl concentration level. Analyses were carried out with HG-AAS. The results show that recoveries were ) 96% in case of the 2-mgrl and 100-grl standards. Recoveries of ) 93% at low concentration levels Ž10 grl. indicated that the method was also well suited to speciation analyses at the trace level. This was particularly important in the case of selenium, where concentrations in the samples were often close to the limit of determination of the method, as the results of speciation studies show. Table 2c contains the results of the reinjection experiments. In the first step, mixtures of 6 mgrl of each ion were injected. After fractionation, an aliquot of 400 l was taken for each ion and reinjected. Consequently, the recovery was quantitative if a total of 80 grl for each ion was found after reinjection. As already mentioned, this was intended to verify the stability of the species during the chromatographic separation, i.e. to prove the absence of redox reactions. This is not always the case. Gerritse and Adeney Ž1985. reported on the oxidation of iodide species during chromatographic separation. They concluded that oxidation was catalyzed by the column material. In case of arsenic, the instability of arsenic ŽIII. had to be taken into consideration. According to Aggett and Kriegman Ž1987., this species is highly susceptible to oxidation to arsenic ŽV., whereas selenium ŽIV. is less prone to oxidation. The outcome of our study gave no indication of instability of either of the species examined. Recoveries of all the reinjected ions were G 95%. Neither arsenic nor selenium species were detected even at the trace level in the other fractions examined. Based on these findings, we considered all the species to be stable during chromatographic separation.
7. Ground waters of the Kelheim region Seven groundwater wells from the Kelheim Region were selected for speciation analyses of arsenic and selenium. Selection was based on the fact that arsenic and selenium concentrations in these samples were sufficiently above the limit of
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
176
Table 2 Quality control experiments: single ions a Ion
Na
Mean wgrlx
S.D. wgrlx
CV Ž%.
Recovery Ž%.
2A As ŽIII. As ŽV. Se ŽIV. Se ŽVI. As ŽIII. As ŽV. Se ŽIV. Se ŽVI. As ŽIII. As ŽV. Se ŽIV. Se ŽVI.
5 5 5 5 5 5 5 5 5 5 5 5
2016 2088 1949 1955 100 102 103 99 9.5 9.3 9.8 9.4
68 67 62 90 8.9 9.7 6.2 12 0.3 0.4 0.5 0.4
3.4 3.2 3.2 4.6 8.9 9.5 6.1 12 3.3 4.5 5.1 4.6
101 104 97 98 100 102 103 99 95 93 98 94
2B Quality control experiments; mixed Ions As ŽIII. 5 2018 As ŽV. 5 2031 Se ŽIV. 5 1929 a Se ŽVI. 5 1981 As ŽIII. 5 99 As ŽV. 5 103 Se ŽIV. 5 101 Se ŽVI. 5 103 As ŽIII. 5 9.6 As ŽV. 5 9.8 Se ŽIV. 5 10.0 Se ŽVI. 5 9.7
96 94 89 78 8.5 10.7 9.3 11.1 0.5 0.8 0.9 0.8
4.7 4.6 4.6 3.9 8.6 10.3 9.2 10.7 5.2 8.1 9.0 8.2
100 101 96 99 99 103 101 103 96 98 100 97
2C Quality control experiments; reinjected ions As ŽIII. 3 5472 As ŽV. 3 5669 Se ŽIV. 3 5740 Se ŽVI. 3 5891 As ŽIII. 3 76 As ŽV. 3 82 Se ŽIV. 3 75 Se ŽVI. 3 76
15 25 35 47 3.0 4.3 1.2 4.0
0.3 0.4 0.6 0.8 3.9 5.3 1.5 5.4
91 95 96 98 95 102 96 95
a
N s Number of runs, S.D.s Standard deviation, CVs Coefficient of variation.
determination of the two elements, i.e. 6.0 grl for arsenic and 5.0 grl for selenium. For this reason, arsenic speciation was carried out in wells A, E, F and G, while selenium speciation was carried out in samples A, B, C and D. Most samples were characterized by elevated concentrations of iron, manganese and sulfur reaching their maximum at the mgrl Žppm. level. Elevated
concentrations were caused by the leaching of pyrites and their oxidation or weathering products, respectively. The variation in the concentrations of Fe, Mn, S, SO42y and DOC in wells A᎐G, as well as their redox potential, pH and conductivity values throughout the sampling seasons are shown in Fig. 1a᎐h. All samples were filtered through a 0.45-m membrane filter prior to chromato-
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
177
Fig. 1 Ža.. Variation of Fe Concentration; Žb. variation of Mn concentration; Žc. variation of S concentration; Žd. variation of SO4 concentration; Že. variation of DOC concentration; Žf. variation of redox potential; Žg. variation of pH; and Žh. variation of conductivity.
graphic separation. Consequently, particulate matter which was defined to have a size ) 0.45 m was removed prior to speciation analyses. All
results presented in this study refer to dissolved species only. The association of arsenic and selenium species with particulate and colloidal matter
178
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
in Kelheim groundwaters was examined by a combination of ultrafiltration and ETV-ICP-MS in a separate study, the results of which are reported elsewhere ŽRaessler et al., 1998.. The concentrations of dissolved iron and manganese in the examined wells varied extremely: maximum values of 4310 grl for iron ŽG, summer. and 4970 grl for manganese ŽB and C, autumn. were contrasted with concentrations of - 5 grl for both elements which were characteristic of well A. Sulfur concentrations in all samples were in the mgrl range with a maximum of 874 mgrl ŽC, autumn.. Additional analyses by ion chromatography showed that sulfur was almost integrally present as sulfate the concentration of which in the wells was between 21 and 1961 mgrl.
Table 3 Speciation of arsenic Sampling season
3A Winter
Spring
Summer
Autumn
Well
As ŽIII. wgrlx
As ŽV. wgrlx
As Žtot. wgrlx
A E F G A E F G A E F G A E F G
n.d. 152.6 73 n.d. n.d. n.d. n.d. n.d. n.d. 176.3 n.d. n.d. n.d. n.d. n.d. n.d.
59.2 n.d. n.d. 35.1 66 114.7 98.4 43.9 70.4 n.d. 104.6 44.9 66.3 147.7 100.2 31.3
59.6 158.0 76.1 38.1 65.3 118.1 105.3 43.9 72.4 179.1 106.7 45.3 65.3 144.7 99.3 33.8
8. Speciation of arsenic
3B Speciation of Selenium
Table 3a indicates that arsenic ŽV. was by far the dominating species in the examined wells. It was the only species detectable in wells A and G, and was also detected in wells E and F during sampling in spring ŽE, F., summer ŽF. and autumn ŽE, F., respectively. On the other hand, As ŽIII. was present in wells E and F in winter ŽE, F. and in summer ŽE.. Similar results for arsenic distribution were found by Hwang and Jiang Ž1994. and Aurillo et al. Ž1994.. Their results, however, refer to arsenic contaminations that are of mostly geogenic origin. Although Mok et al. Ž1988. reported on arsenic in groundwaters near a mining area of the United States and Wennrich Ž1997. examined arsenic speciation of a percolate water from a tin mill tailing in Germany, there is still a need for more detailed information on the distribution of arsenic species in anthropogenically contaminated water samples. Undoubtedly, the predominance of As ŽV. in our samples was a consequence of the roasting of arsenic-containing pyrites, which can be summarized as follows:
Sampling season
Well
Se ŽIV. wgrlx
Se ŽVI. wgrlx
Se Žtot. wgrlx
Winter
A B C D A B C D A B C D A B C D
n.d. n.d. n.d. 6,0 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 5.0 19.7
12.8 328 60.9 33.1 19.1 40.6 50.9 16.5 18.7 52.1 13.5 n.d. 22.6 89 10.5 n.d.
12.5 325.6 61.9 40.9 21.6 44.5 52.7 13.7 19.9 54.1 12.9 n.d. 25.7 89.4 17.3 23.5
4FeAsSq 13O2 q 6H2 O ª 4FeSO4 q 4H3 AsO4
Spring
Summer
Autumn
i.e. arsenic was leached as As ŽV. and penetrated the groundwater. Moreover, the predominance of arsenic ŽV. was favored thermodynamically and by the higher ionic strength of the solution. This agrees with the observations made in our samples. Positive redox potentials were measured in all wells favoring the presence of As ŽV., with the exception of well E during the winter and summer sampling. In the latter cases, negative redox
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
potentials of y24.5 and y20.5 mV, respectively, were measured. The only detectable species was As ŽIII.. As ŽIII. was also detected in well F during winter sampling. In this case, however, there was a positive redox potential of 60 mV. Consequently, the presence of As ŽIII. was not only a consequence of thermodynamic parameters. As ŽIII. was probably favored ᎏ or at least stabilized ᎏ by interaction with organic matter. This assumption was also made by Yokoyama et al. Ž1993.. Moreover, Van Elteren et al. Ž1991. observed the complete reduction of arsenic ŽV. to arsenic ŽIII. in deionized water, possibly due to traces of organic matter from ion exchange cartridges used for demineralization. There were elevated DOC concentrations in well E Žwinter: 41.4 mgrl; summer 37.9 mgrl.. In case of well F, the DOC concentration in winter was 22.9 mgrl. These concentrations were considerably higher than in most other samples. Elevated DOC concentrations in well E could be traced to its location in the grounds of a former cellulose-producing plant. Although the structure of the organic matter in these samples has not yet been identified in detail, there was good reason to assume that, apart from humic substances, considerable amounts of lignine sulfonic acids are present. On the other hand, the coincidence of smaller DOC concentrations and positive redox potentials resulted in the presence of arsenic ŽV. in wells E and F, as was the case for springrautumn ŽE. and for all sampling seasons, except winter in the case of F. On the other hand, long-term monitoring of species in wells A and G, however, revealed an almost homogeneous distribution of arsenic ŽV. concentration throughout the year, which was also true for well F except for winter sampling. The concentration of total arsenic in the wells seemed to be almost independent of the concentration of both dissolved iron and manganese. As shown by well F, the concentration of arsenic remained almost unchanged during spring, summer and autumn, whereas the concentration of dissolved iron and manganese in the well dropped constantly during this period. Nonetheless, it is important to note that distribution of arsenic species in sample E changed completely between the respective sampling seasons. If there had been
179
only one sampling and speciation analysis of this well, the variation of species distribution throughout the year could not have been revealed.
9. Speciation of selenium Table 3b indicates that selenium ŽVI. was by far the dominating species in the examined wells. It was the only species detected in wells A and B throughout the year, and in well C, except for autumn sampling. In well D the selenium speciation turned out to be very complex: the common presence of both selenium ŽIV. and selenium ŽVI. in winter; only selenium ŽVI. in spring; no selenium at all detectable in summer; and only selenium ŽIV. in autumn sampling. Selenium speciation turned out to be more difficult than that of arsenic since the concentration of selenium ŽIV. was close to or even at the limit of determination in wells C Žautumn. and D Žwinter.. In some cases, e.g. well B Žspring., traces of selenium ŽIV. were presumably present but were not amenable to analysis. Selenium ŽIV., however, was unequivocally determined in well D Žautumn. where its concentration was 19.7 grl and, therefore, sufficiently above the limit of determination of the method Ž5.0 grl.. The predominance of Se ŽVI. in groundwater samples was also observed by Tanzer and Heumann Ž1991., Alfthan et al. Ž1995. and Reddy et al. Ž1995.. In this context the question is discussed whether the predominance of Se ŽVI. can be explained by the precipitation of Se ŽIV. due to the lower solubility of the latter species. Moreover, selenite was reported by An and Langqiu Ž1987. to be strongly retained by iron and manganese oxyhydroxides. In fact, most of our samples where selenium ŽVI. was the predominant species were characterized by elevated concentrations of both iron and manganese of up to nearly 5 mgrl Žwells B and C, autumn.. Well D Žautumn., with a unique presence of selenium ŽIV., contained only little iron and manganese. Thermodynamic parameters, such as the redox potential, seemed to be of only little influence on selenium speciation. All wells analyzed throughout the year had positive redox potentials, irrespective of the
180
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
selenium species present. Although DOC concentrations were in general considerably lower than in wells E, F and G, it seemed likely that there was an interaction between organic matter and selenium ŽIV.. This could be demonstrated by size fractionation studies of selenium in wells with only little dissolved iron and manganese ŽRaessler et al., 1998.. Despite the lack of both iron and manganese, colloidal fractions of selenium were detected and attributed to an association with organic matter. Nonetheless, further experiments will be necessary to unequivocally answer the question of whether the presence of selenium ŽIV. is be favored by humic or lignine sulfonic acids. If both selenium ŽIV. and selenium ŽVI. were present, part of the selenite in our samples was between 14᎐35% and was, therefore, slightly higher than that observed by Tanzer and Heumann Ž1991. who indicated a value of - 10%. On the contrary, in Finnish groundwaters analyzed by Alfthan et al. Ž1995., selenite ranged from 6᎐36% and was, therefore, close to the values we found in our samples.
10. Conclusion Both the concentration and distribution of arsenic and selenium species in groundwaters from Kelheim were monitored throughout the year by a combination of HPLC and HG-AAS. The presented method turned out to be reliable, rugged and well-suited to determining the concentration of both arsenic and selenium species with respect to the threshold value of 10 grl indicated by German drinking water regulation. The reliability of the method was proven by a number of quality control experiments. The longterm application of the method to groundwaters from Kelheim showed the predominance of arsenic ŽV. and selenium ŽVI. in most of the samples, which was in agreement with literature. Samples were taken from seven wells in winter, spring, summer and autumn. With the exception of well E, the distribution and concentration of arsenic species in most samples was almost homogenous throughout the year. Well E, how-
ever, was characterized by a complete change of species distribution between the sample seasons, which emphasized the necessity of long-term monitoring studies like this one to obtain reliable data. These are indispensable to take appropriate measures, e.g. in groundwater protection. In the majority of samples, arsenic was present as As ŽV. except for samples that were characterized by a negative redox potential andror high amounts of organic matter. In case of selenium, considerable variation of distribution of species was observed. While the presence of both arsenic ŽIII. and ŽV. in the Kelheim groundwaters was obviously favored by thermodynamic parameters, this was not the case for selenium ŽIV. whose existence seemed to profit from minor concentrations of iron and manganese in the wells. Further investigations will be needed to examine the interaction of arsenic and selenium species with organic matter. There is good reason to assume that presence of humic and lignine sulfonic acids will support the existence of arsenic ŽIII. and selenium ŽIV. in groundwater samples.
Acknowledgements Michael Raessler thanks Prof K.P. Seiler, GSF National Research Centre of Environment and Health, Institute of Hydrology, for generous financial support to carry out this study. References Aggett J, Kriegman MR. Preservation of arsenic ŽIII. and arsenic ŽV. in sample sediment interstitial water. Analyst 1987;112:153᎐157. Alfthan G, Wang D, Aro A, Soveri J. The geochemistry of selenium in groundwaters in Finland. Sci Total Environ 1995;162:93᎐103. An P, Langqiu X. The effects of humic acid on the chemical and biological properties of selenium in the environment. Sci Total Environ 1987;64:89᎐98. Aurillo AC, Mason RP, Hemond HF. Speciation and fate of arsenic in three lakes of the Aberjona watershed. Environ Sci Technol 1994;28:577᎐583. Gerritse RG, Adeney JA. Rapid determination in water of chloride, sulphate, sulphite, selenite, selenate and arsenate among other inorganic and organic solutes by ion chro-
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181 matography with UV detection below 195 nm. J Chromatogr A 1985;347:419᎐428. Hwang CJ, Jiang SJ. Determination of arsenic compounds in water samples by liquid chromatography-inductively coupled mass spectrometry with an in situ nebulizer-hydride generator. Anal Chim Acta 1994;289:205᎐213. Michalke B. Quality control and reference materials in speciation analysis. Fresenius J Anal Chem 1999;363:439᎐445. Mok WM, Riley JA, Wai CM. Arsenic speciation and quality of groundwater in a lead᎐zinc mine, Idaho. Water Res 1988;22:769᎐774. Raessler M, Michalke B, Schramel P, Schulte-Hostede S, Kettrup A. The capability of ultrafiltration and ETV-ICPMS for size fractionation studies of arsenic and selenium species in ground water samples with high concentrations of iron, manganese and sulfur. Fresenius J Anal Chem 1998;362:281᎐284. Raessler M. Ph.D. Thesis. Technical University Munich, Germany, 1997. Reddy KJ, Zhang Z, Blaylock MJ, Vance GF. Method for detecting selenium speciation in groundwater. Environ Sci Technol 1995;29:1754᎐1759.
181
Smith AH, Hoppenheyn-Rich C, Bates MN et al. Cancer risks from arsenic in drinking water. Environ Health Persp 1992;97:259᎐267. Tanzer D, Heumann KG. Determination of selenium species in environmental water samples using isotope dilution mass spectrometry. Anal Chem 1991;63:1984᎐1989. Van Elteren JT, Hoegee J, van der Hoek EE, Das HA, de Ligny CL, Agterdenbos J. Preservation of arsenic ŽIII. and arsenic ŽV. in some water samples. J Radioanal Nucl Chem 1991;154:343᎐347. Valentine JL, Cebrian ME, Garcia-Vargas GG et al. Daily selenium intake estimates for residents of arsenic-endemic areas. Environ Res 1994;64:1᎐9. Verordnung ueber Trinkwasser und ueber Wasser fuer Lebensmittelbetriebe. Trinkwasserverordnung 1996. German drinking water regulation, 1996. Wennrich R. Determination of arsenic species in water samples of a tin ore tailing. Vom Wasser 1997;88:1᎐12. Yokoyama T, Takahashi Y, Tarutani T. Simultaneous determination of arsenic and arsenious acids in geothermal water. Chem Geol 1993;103:103᎐111.
Long-term monitoring of arsenic and selenium species in contaminated groundwaters by HPLC and HG-AAS Michael Raessler U , Bernhard Michalke, Sigurd Schulte-Hostede, Antonius Kettrup GSF National Research Centre for En¨ ironment and Health, Institute of Eco-logical Chemistry, P.O. Box 1129, D-85758, Neuherberg, Germany Received 15 January 2000; accepted 12 May 2000
Abstract The long-term concentration and distribution of species of arsenic and selenium in contaminated groundwaters from Kelheim was monitored. Most of the groundwater wells contained elevated concentrations of iron, manganese and sulfur. Arsenic ŽIII., arsenic ŽV., selenium ŽIV. and selenium ŽVI. were separated using high performance liquid chromatography ŽHPLC. based on phosphate buffers and collected in fractions. Due to the complex matrix, the fractions were analyzed element-specifically by hydride-generating atomic absorption spectrometry ŽHG-AAS.. The combination of HPLC and HG-AAS was selected due to the authors’ intention of developing an easy-to-handle, but nonetheless reliable, method suitable for the long-term monitoring of species distribution in an almost routine way, and taking account of the threshold values of 10 grl for each element, indicated by German drinking water regulations. To enhance the reliability of the method, analytical quality control experiments were carried out. When applied to groundwater wells from Kelheim ŽGermany. they revealed that arsenic ŽV. and selenium ŽVI. were the dominating species. The presence of arsenic ŽIII. and selenium ŽIV. was assumed to be supported by organic matter. 䊚 2000 Elsevier Science B.V. All rights reserved. Keywords: Arsenic; Selenium; Speciation; Long-term monitoring; Quality control
U
Corresponding author. Max-Planck-Institute for Biogeochemistry, Tatzendpromenade 1A, D-07745 Jena, Germany. Tel.: q49-3641-64-3812; fax:q 49-3641-64-3710. E-mail address: [email protected] ŽM. Raessler.. 0048-9697r00r$ - see front matter 䊚 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 8 - 9 6 9 7 Ž 0 0 . 0 0 5 3 5 - 0
172
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
1. Introduction Arsenic and selenium may have toxic effects on biota in groundwater samples. Humans may be affected if the concentrations are high. For this reason, the German drinking water regulation ŽVerordnung ueber Trinkwasser, 1996. indicates a threshold value of 10 grl for each element. Arsenic is supposed to induce cancer ŽSmith et al., 1992., and elevated concentrations of selenium may lead to selenosis ŽValentine et al., 1994.. However, it is known today that the toxicity of an element is determined by the chemical form in which it is present in the sample. For example, organic arsenobetaine, the most common arsenic compound in fish, is non-toxic to humans, whereas arsenic ŽIII. is more toxic than arsenic ŽV.. In case of selenium, selenium ŽIV. is considered more dangerous to aquatic organisms than selenium ŽVI. due to its higher solubility and bioavailabilty. Consequently, for adequate risk assessment, modern environmental research must be no longer restricted to the determination of the total concentration of an element, but must account for the different ‘species’ of this element present in the sample. It is the aim of speciation to reveal the identity and the quantity of different specific compounds or oxidation states of an element. For speciation to be quantitative, the sum of the concentration of the element in all species present must be equal to the total concentration of the element in the sample. Unfortunately, in most cases, speciation is still restricted to a relatively small number of very specialized research laboratories. Its application to the monitoring of groundwater, for example, is still far from being routine work. It was the authors’ intention to develop a simple, rugged and, nonetheless, reliable method for the speciation of inorganic arsenic and selenium and to apply it to groundwater wells from Kelheim, Germany. The samples were characterized by high concentrations of iron, manganese and sulfur due to leaching processes of pyrites and their oxidation products, iron oxides. Pyrites were formerly used for the production of sulfuric acid. Both pyrites and iron oxides were deposited in the Kelheim region or were
used as filling materials in road construction and railway embankments. Among other trace elements, pyrites and iron oxides contained elevated concentrations of leachable arsenic and selenium that penetrated into the groundwater. High concentrations of interfering ions in these samples made the direct determination of arsenic and selenium by ion chromatography based on conductivity or UV detection impossible. For this reason, a combination of HPLC and elementspecific detection was mandatory. HG-AAS was selected for the following reasons: Ž1. its high selectivity towards arsenic and selenium; Ž2. the fact that the determination of arsenic and selenium is almost free of matrix interferences since hydride-forming elements ŽAs, Se. are separated from the matrix prior to atomization; and Ž3. HG-AAS, which offers detection limits for arsenic and selenium of - 1 grl, which meet the needs for most environmental applications, is much less expensive than the more sophisticated ICP-MS. As the results of a reliability study and of analyses of Kelheim groundwaters show, the combination of HPLC and HG-AAS is well suited for the species-monitoring of arsenic and selenium in environmental samples.
2. Experimental section 2.1. Reagents The chemicals used throughout this procedure were of analytical grade or better. Deionized water was produced with Millipore deionizing cartridges. The mobile phase was prepared by dissolving 0.575 g NH4 H2 PO4 ŽAldrich., 1.32 g ŽNH4 .2 HPO4 ŽAldrich. and 2.13 g Na2 SO4 ŽFluka. in deionized water. The addition of Na2 SO4 was necessary for the quantitative recovery of selenate. This solution was transferred to a 1-l flask and filled to the mark. The pH of the eluent was adjusted to 5.9 by adding, dropwise, H3 PO4 85% ŽFluka.. Prior to use, the eluent was filtered through a 450-nm filter ŽMillipore. and degassed in an ultrasonic bath for 1 h. Stock solutions Žvolumes 100 ml, concentration s 1 grl, i.e. 1000 ppm. of arsenate, selenite
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
173
Table 1 Chromatographic system and operating parameters 1A Pump Injection valve Sample loop Column Žsize. Eluent Flow rate UV detector Wavelength Plotter Fraction collector
Beckman 114 M solvent delivery module Beckman 7725i ŽRheodyne. 100 l Hamilton PRP X-100 Ž250 = 4.6 mm. 5 mM ADP, 10 mM DAP, 15 mM Na2SO4, pH 5.9 1 mlrmin BioRad Model 1706 UVrVIS monitor s 195 nm Spectra Physics SP 4270 integrator BioRad Model 2110 fraction collector
1B Parameters for HG-AAS detection Model Hydride generator Cell temperature Excitation source Wavelength Purge I Purge II Reaction time Slit width
Perkin-Elmer 3030B MHS 20 1000⬚C EDL 193.7 nm ŽAs.; 196.0 nm ŽSe. 50 s ŽAs.; 20 s ŽSe. 40 s ŽAs.; 40 s ŽSe. 18 s ŽAs.; 12 s ŽSe. 0.7 nm
and selenate were prepared by dissolving 0.416 g Na2 HAsO4 ⭈ 7H2 O ŽFluka., 0.219 g Na2 SeO3 ŽAldrich. and 0.467 g Na2 SeO4 ⭈ 10H2 O ŽAldrich., respectively, in water. The stock solution of As ŽIII. was prepared by dissolving 0.132 g As2 O3 ŽAldrich. in 5 ml of 1 M NaOH and filling to the mark with water. Single ions and mixed ion wAs ŽIII., As ŽV., Se ŽIV. q Se ŽVI.x working standards were prepared prior to use by adequate dilution of the stock solutions. 2.2. Chromatographic separation The separation of species was carried out by HPLC. The chromatographic system and operating parameters are listed in Table 1a. Three column materials were tested for their ability of species separation: Nucleosil 100-5 NŽCH3 .2 ET 250r4 ŽMacherey and Nagel., Vydac 302 IC 4.6, 250 = 4.6 mm Žthe Separations Group. and Hamilton PRP X-100 ŽHamilton Corporation.. Whereas the stationary phase of the first two columns were silica-based, the stationary phase of the Hamilton column was made of a
polystyrol-divinylbenzene matrix which was stable in a broad pH range. Of the three columns tested, the Hamilton column showed the best separation efficiency under isocratic conditions and was, therefore, used for further experiments with model solutions and speciation analyses of groundwater samples. Retention times were as follows: arsenic ŽIII. 2.6 min; arsenic ŽV.; selenium ŽIV. 4.0 min; and selenium ŽVI. 11.1 min. The coelution of arsenic ŽV. and selenium ŽIV. was considered to be no drawback due to element-specific detection. For the quantitative recovery of selenate the ionic strength of the phosphate buffer eluent had to be enhanced by addition of 15 mM Na2 SO4 . Na2 SO4 was selected since it did not influence the pH of the eluent nor UV detection Žno signal at 195 nm.. UV detection was used for method development only. Without addition of sulfate, the recovery of selenate was - 80%. The overall separation lasted 15 min. Prior to quantification, the ions were collected in fractions. Five fractions were taken every 3 min. If present, As ŽIII. was found in fraction 1 Ž0᎐3
174
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
min., As ŽV. andror Se ŽIV. in fraction 2 Ž3᎐6 min. and Se ŽVI. in fraction 4 Ž9᎐12 min.. Fractions 3 Ž6᎐9 min. and 5 Ž12᎐15 min. did not contain either arsenic or selenium species. The collected fractions were immediately analyzed by HG-AAS.
3. Element-specific detection Instrumentation and operating parameters for the detection of species by HG-AAS are listed in Table 1b. A solution of 3% NaBH4 was used for hydride generation. The volume of each fraction was extended to 10 ml by adding 7 ml of 3% HCl ŽMerck.. Since all the arsenic or selenium in the sample was purged into the atomizer, addition of HCl was not a dilution of the sample. Standards and samples containing Se ŽVI. were reduced to Se ŽIV. prior to analysis by HG-AAS since Se ŽVI. did not form a hydride under the abovementioned conditions. A quantitative reduction was carried out by adding 5 ml of HCl 32% to 5 ml of the sample in Nalgene tubes. The tubes were sealed and exposed to a water bath at 80⬚C for 1 h. After reduction, the samples were cooled to room temperature and analyzed. The analyses were evaluated based upon calibration curves in the range 1.5᎐15 grl for each element. The detection limits of HG-AAS for arsenic and selenium were based upon the analyses of 10 blanks of HCl suprapur 3% ŽMerck.. The corresponding values were 0.6 grl ŽAs. and 0.5 grl ŽSe. with t Ž f s 9, Ps 95%, one-sided.. A 100-l sample loop was used for HPLC Žcf Table 1a.. The limit of determination of the overall method was given by the combination of both the detection limit of HG-AAS and the volume of the sample loop. Consequently, this was found to be 6 grl for arsenic and 5 grl for selenium. The total concentrations of arsenic and selenium were determined by both HG-AAS and ETV-ICP-MS to confirm the reliability of HGAAS. For the exact working conditions of ETVICP-MS, see Raessler et al. Ž1998.. A JY-70 ICP atomic emission spectrometer ŽJobin Yvon Instruments SA. was used to determine the Fe, Mn
and S concentrations in the groundwater samples. Analyses were carried out at s 259.9 nm ŽFe., 257.6 nm ŽMn. and 181.9 nm ŽS.. The concentration of SO42y was determined by ion chromatography ŽDionex DX 500.. The determination of DOC was carried out in the filtered samples using a high TOC analyzer ŽElementar. while the redox potential, pH and conductivity were determined with a WTW Multical instrument.
4. Quality control Model solutions containing single ions of As ŽIII., As ŽV., Se ŽIV. and Se ŽVI., as well as mixed standard solutions of all ions, were used to check the reliability of the chromatographic separation and the element-specific detection. Recoveries of the ions at concentration levels of 2 mgrl, 100 and 10 grl were determined and mass balances established for each ion. Besides the quantitative recovery, the stability of all species was evaluated by what we called ‘reinjection experiments’. Referring to arsenic ŽIII., reinjection was carried out as follows: the chromatographic separation and fractionation were carried out as described above. After the quantitative recovery of arsenic ŽIII. in fraction 1, an aliquot of this fraction was reinjected and the fractionation repeated. All five fractions were, once again, analyzed by HG-AAS. The reinjection experiment was successful, i.e. arsenic ŽIII. was assumed to be stable during the chromatographic separation if its recovery in fraction 1 was again quantitative and if there was no arsenic present in the other fractions. Analogous experiments were carried out for arsenic ŽV., selenium ŽIV. and selenium ŽVI..
5. Collection and preparation of groundwaters of the Kelheim region Groundwater samples were collected from seven different groundwater wells ŽA᎐G. at Kelheim, Lower Bavaria, Germany. The city of Kelheim is situated at the confluence of Altmuhl ¨ river and Danube river in the south-east of Germany. The sampling points were located in the
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
grounds of a chemical plant where sulfuric acid was produced; in the grounds of a cellulose-producing plant that had been shut down, and in the city center of Kelheim. Samples were taken during the winter, spring, summer and autumn seasons to allow monitoring of species throughout the year. Samples were pumped with a delivery of 1.2 lrs. To avoid contamination by the tubing, the pumping system was conditioned with well water. The first 50 l of each sample was discarded. Samples were filled into 1-l PE bottles, and stored at 4⬚C, protected from daylight. To avoid the formation and precipitation of iron and manganese oxyhydroxides, the samples were acidified to pH- 2 with HCl. The filled bottles were immediately sealed to avoid exposure to air. To avoid plugging of the HPLC valves, all samples were vacuum filtered through 0.45-m filtration membranes prior to HPLC separation. The separation and quantitation of species were carried out as described above.
6. Results and discussion 6.1. Quality control Quality control is of growing importance to speciation analyses ŽMichalke, 1999. which, in most cases, still suffer from a lack of adequate standard reference materials. The accuracy of results of speciation analyses is influenced by various parameters, such as sampling, sample storage and sample preparation. The separation of species and their analytical quantitation are the final steps of the speciation procedure. While it is almost impossible not to influence the distribution of species at least partly during the sampling procedure, every effort should be made to verify that there is no detrimental influence on the species that could be traced to separation and quantitation. This was the motivation for the reliability study results, which are presented here. More details referring to the quality control of this procedure are given by Raessler Ž1997.. Table 2a shows the results for the injection of single ions; and Table 2b the results for mixed ion
175
solutions at the 2 mgrl, 100 and 10 grl concentration level. Analyses were carried out with HG-AAS. The results show that recoveries were ) 96% in case of the 2-mgrl and 100-grl standards. Recoveries of ) 93% at low concentration levels Ž10 grl. indicated that the method was also well suited to speciation analyses at the trace level. This was particularly important in the case of selenium, where concentrations in the samples were often close to the limit of determination of the method, as the results of speciation studies show. Table 2c contains the results of the reinjection experiments. In the first step, mixtures of 6 mgrl of each ion were injected. After fractionation, an aliquot of 400 l was taken for each ion and reinjected. Consequently, the recovery was quantitative if a total of 80 grl for each ion was found after reinjection. As already mentioned, this was intended to verify the stability of the species during the chromatographic separation, i.e. to prove the absence of redox reactions. This is not always the case. Gerritse and Adeney Ž1985. reported on the oxidation of iodide species during chromatographic separation. They concluded that oxidation was catalyzed by the column material. In case of arsenic, the instability of arsenic ŽIII. had to be taken into consideration. According to Aggett and Kriegman Ž1987., this species is highly susceptible to oxidation to arsenic ŽV., whereas selenium ŽIV. is less prone to oxidation. The outcome of our study gave no indication of instability of either of the species examined. Recoveries of all the reinjected ions were G 95%. Neither arsenic nor selenium species were detected even at the trace level in the other fractions examined. Based on these findings, we considered all the species to be stable during chromatographic separation.
7. Ground waters of the Kelheim region Seven groundwater wells from the Kelheim Region were selected for speciation analyses of arsenic and selenium. Selection was based on the fact that arsenic and selenium concentrations in these samples were sufficiently above the limit of
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
176
Table 2 Quality control experiments: single ions a Ion
Na
Mean wgrlx
S.D. wgrlx
CV Ž%.
Recovery Ž%.
2A As ŽIII. As ŽV. Se ŽIV. Se ŽVI. As ŽIII. As ŽV. Se ŽIV. Se ŽVI. As ŽIII. As ŽV. Se ŽIV. Se ŽVI.
5 5 5 5 5 5 5 5 5 5 5 5
2016 2088 1949 1955 100 102 103 99 9.5 9.3 9.8 9.4
68 67 62 90 8.9 9.7 6.2 12 0.3 0.4 0.5 0.4
3.4 3.2 3.2 4.6 8.9 9.5 6.1 12 3.3 4.5 5.1 4.6
101 104 97 98 100 102 103 99 95 93 98 94
2B Quality control experiments; mixed Ions As ŽIII. 5 2018 As ŽV. 5 2031 Se ŽIV. 5 1929 a Se ŽVI. 5 1981 As ŽIII. 5 99 As ŽV. 5 103 Se ŽIV. 5 101 Se ŽVI. 5 103 As ŽIII. 5 9.6 As ŽV. 5 9.8 Se ŽIV. 5 10.0 Se ŽVI. 5 9.7
96 94 89 78 8.5 10.7 9.3 11.1 0.5 0.8 0.9 0.8
4.7 4.6 4.6 3.9 8.6 10.3 9.2 10.7 5.2 8.1 9.0 8.2
100 101 96 99 99 103 101 103 96 98 100 97
2C Quality control experiments; reinjected ions As ŽIII. 3 5472 As ŽV. 3 5669 Se ŽIV. 3 5740 Se ŽVI. 3 5891 As ŽIII. 3 76 As ŽV. 3 82 Se ŽIV. 3 75 Se ŽVI. 3 76
15 25 35 47 3.0 4.3 1.2 4.0
0.3 0.4 0.6 0.8 3.9 5.3 1.5 5.4
91 95 96 98 95 102 96 95
a
N s Number of runs, S.D.s Standard deviation, CVs Coefficient of variation.
determination of the two elements, i.e. 6.0 grl for arsenic and 5.0 grl for selenium. For this reason, arsenic speciation was carried out in wells A, E, F and G, while selenium speciation was carried out in samples A, B, C and D. Most samples were characterized by elevated concentrations of iron, manganese and sulfur reaching their maximum at the mgrl Žppm. level. Elevated
concentrations were caused by the leaching of pyrites and their oxidation or weathering products, respectively. The variation in the concentrations of Fe, Mn, S, SO42y and DOC in wells A᎐G, as well as their redox potential, pH and conductivity values throughout the sampling seasons are shown in Fig. 1a᎐h. All samples were filtered through a 0.45-m membrane filter prior to chromato-
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
177
Fig. 1 Ža.. Variation of Fe Concentration; Žb. variation of Mn concentration; Žc. variation of S concentration; Žd. variation of SO4 concentration; Že. variation of DOC concentration; Žf. variation of redox potential; Žg. variation of pH; and Žh. variation of conductivity.
graphic separation. Consequently, particulate matter which was defined to have a size ) 0.45 m was removed prior to speciation analyses. All
results presented in this study refer to dissolved species only. The association of arsenic and selenium species with particulate and colloidal matter
178
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
in Kelheim groundwaters was examined by a combination of ultrafiltration and ETV-ICP-MS in a separate study, the results of which are reported elsewhere ŽRaessler et al., 1998.. The concentrations of dissolved iron and manganese in the examined wells varied extremely: maximum values of 4310 grl for iron ŽG, summer. and 4970 grl for manganese ŽB and C, autumn. were contrasted with concentrations of - 5 grl for both elements which were characteristic of well A. Sulfur concentrations in all samples were in the mgrl range with a maximum of 874 mgrl ŽC, autumn.. Additional analyses by ion chromatography showed that sulfur was almost integrally present as sulfate the concentration of which in the wells was between 21 and 1961 mgrl.
Table 3 Speciation of arsenic Sampling season
3A Winter
Spring
Summer
Autumn
Well
As ŽIII. wgrlx
As ŽV. wgrlx
As Žtot. wgrlx
A E F G A E F G A E F G A E F G
n.d. 152.6 73 n.d. n.d. n.d. n.d. n.d. n.d. 176.3 n.d. n.d. n.d. n.d. n.d. n.d.
59.2 n.d. n.d. 35.1 66 114.7 98.4 43.9 70.4 n.d. 104.6 44.9 66.3 147.7 100.2 31.3
59.6 158.0 76.1 38.1 65.3 118.1 105.3 43.9 72.4 179.1 106.7 45.3 65.3 144.7 99.3 33.8
8. Speciation of arsenic
3B Speciation of Selenium
Table 3a indicates that arsenic ŽV. was by far the dominating species in the examined wells. It was the only species detectable in wells A and G, and was also detected in wells E and F during sampling in spring ŽE, F., summer ŽF. and autumn ŽE, F., respectively. On the other hand, As ŽIII. was present in wells E and F in winter ŽE, F. and in summer ŽE.. Similar results for arsenic distribution were found by Hwang and Jiang Ž1994. and Aurillo et al. Ž1994.. Their results, however, refer to arsenic contaminations that are of mostly geogenic origin. Although Mok et al. Ž1988. reported on arsenic in groundwaters near a mining area of the United States and Wennrich Ž1997. examined arsenic speciation of a percolate water from a tin mill tailing in Germany, there is still a need for more detailed information on the distribution of arsenic species in anthropogenically contaminated water samples. Undoubtedly, the predominance of As ŽV. in our samples was a consequence of the roasting of arsenic-containing pyrites, which can be summarized as follows:
Sampling season
Well
Se ŽIV. wgrlx
Se ŽVI. wgrlx
Se Žtot. wgrlx
Winter
A B C D A B C D A B C D A B C D
n.d. n.d. n.d. 6,0 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 5.0 19.7
12.8 328 60.9 33.1 19.1 40.6 50.9 16.5 18.7 52.1 13.5 n.d. 22.6 89 10.5 n.d.
12.5 325.6 61.9 40.9 21.6 44.5 52.7 13.7 19.9 54.1 12.9 n.d. 25.7 89.4 17.3 23.5
4FeAsSq 13O2 q 6H2 O ª 4FeSO4 q 4H3 AsO4
Spring
Summer
Autumn
i.e. arsenic was leached as As ŽV. and penetrated the groundwater. Moreover, the predominance of arsenic ŽV. was favored thermodynamically and by the higher ionic strength of the solution. This agrees with the observations made in our samples. Positive redox potentials were measured in all wells favoring the presence of As ŽV., with the exception of well E during the winter and summer sampling. In the latter cases, negative redox
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
potentials of y24.5 and y20.5 mV, respectively, were measured. The only detectable species was As ŽIII.. As ŽIII. was also detected in well F during winter sampling. In this case, however, there was a positive redox potential of 60 mV. Consequently, the presence of As ŽIII. was not only a consequence of thermodynamic parameters. As ŽIII. was probably favored ᎏ or at least stabilized ᎏ by interaction with organic matter. This assumption was also made by Yokoyama et al. Ž1993.. Moreover, Van Elteren et al. Ž1991. observed the complete reduction of arsenic ŽV. to arsenic ŽIII. in deionized water, possibly due to traces of organic matter from ion exchange cartridges used for demineralization. There were elevated DOC concentrations in well E Žwinter: 41.4 mgrl; summer 37.9 mgrl.. In case of well F, the DOC concentration in winter was 22.9 mgrl. These concentrations were considerably higher than in most other samples. Elevated DOC concentrations in well E could be traced to its location in the grounds of a former cellulose-producing plant. Although the structure of the organic matter in these samples has not yet been identified in detail, there was good reason to assume that, apart from humic substances, considerable amounts of lignine sulfonic acids are present. On the other hand, the coincidence of smaller DOC concentrations and positive redox potentials resulted in the presence of arsenic ŽV. in wells E and F, as was the case for springrautumn ŽE. and for all sampling seasons, except winter in the case of F. On the other hand, long-term monitoring of species in wells A and G, however, revealed an almost homogeneous distribution of arsenic ŽV. concentration throughout the year, which was also true for well F except for winter sampling. The concentration of total arsenic in the wells seemed to be almost independent of the concentration of both dissolved iron and manganese. As shown by well F, the concentration of arsenic remained almost unchanged during spring, summer and autumn, whereas the concentration of dissolved iron and manganese in the well dropped constantly during this period. Nonetheless, it is important to note that distribution of arsenic species in sample E changed completely between the respective sampling seasons. If there had been
179
only one sampling and speciation analysis of this well, the variation of species distribution throughout the year could not have been revealed.
9. Speciation of selenium Table 3b indicates that selenium ŽVI. was by far the dominating species in the examined wells. It was the only species detected in wells A and B throughout the year, and in well C, except for autumn sampling. In well D the selenium speciation turned out to be very complex: the common presence of both selenium ŽIV. and selenium ŽVI. in winter; only selenium ŽVI. in spring; no selenium at all detectable in summer; and only selenium ŽIV. in autumn sampling. Selenium speciation turned out to be more difficult than that of arsenic since the concentration of selenium ŽIV. was close to or even at the limit of determination in wells C Žautumn. and D Žwinter.. In some cases, e.g. well B Žspring., traces of selenium ŽIV. were presumably present but were not amenable to analysis. Selenium ŽIV., however, was unequivocally determined in well D Žautumn. where its concentration was 19.7 grl and, therefore, sufficiently above the limit of determination of the method Ž5.0 grl.. The predominance of Se ŽVI. in groundwater samples was also observed by Tanzer and Heumann Ž1991., Alfthan et al. Ž1995. and Reddy et al. Ž1995.. In this context the question is discussed whether the predominance of Se ŽVI. can be explained by the precipitation of Se ŽIV. due to the lower solubility of the latter species. Moreover, selenite was reported by An and Langqiu Ž1987. to be strongly retained by iron and manganese oxyhydroxides. In fact, most of our samples where selenium ŽVI. was the predominant species were characterized by elevated concentrations of both iron and manganese of up to nearly 5 mgrl Žwells B and C, autumn.. Well D Žautumn., with a unique presence of selenium ŽIV., contained only little iron and manganese. Thermodynamic parameters, such as the redox potential, seemed to be of only little influence on selenium speciation. All wells analyzed throughout the year had positive redox potentials, irrespective of the
180
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181
selenium species present. Although DOC concentrations were in general considerably lower than in wells E, F and G, it seemed likely that there was an interaction between organic matter and selenium ŽIV.. This could be demonstrated by size fractionation studies of selenium in wells with only little dissolved iron and manganese ŽRaessler et al., 1998.. Despite the lack of both iron and manganese, colloidal fractions of selenium were detected and attributed to an association with organic matter. Nonetheless, further experiments will be necessary to unequivocally answer the question of whether the presence of selenium ŽIV. is be favored by humic or lignine sulfonic acids. If both selenium ŽIV. and selenium ŽVI. were present, part of the selenite in our samples was between 14᎐35% and was, therefore, slightly higher than that observed by Tanzer and Heumann Ž1991. who indicated a value of - 10%. On the contrary, in Finnish groundwaters analyzed by Alfthan et al. Ž1995., selenite ranged from 6᎐36% and was, therefore, close to the values we found in our samples.
10. Conclusion Both the concentration and distribution of arsenic and selenium species in groundwaters from Kelheim were monitored throughout the year by a combination of HPLC and HG-AAS. The presented method turned out to be reliable, rugged and well-suited to determining the concentration of both arsenic and selenium species with respect to the threshold value of 10 grl indicated by German drinking water regulation. The reliability of the method was proven by a number of quality control experiments. The longterm application of the method to groundwaters from Kelheim showed the predominance of arsenic ŽV. and selenium ŽVI. in most of the samples, which was in agreement with literature. Samples were taken from seven wells in winter, spring, summer and autumn. With the exception of well E, the distribution and concentration of arsenic species in most samples was almost homogenous throughout the year. Well E, how-
ever, was characterized by a complete change of species distribution between the sample seasons, which emphasized the necessity of long-term monitoring studies like this one to obtain reliable data. These are indispensable to take appropriate measures, e.g. in groundwater protection. In the majority of samples, arsenic was present as As ŽV. except for samples that were characterized by a negative redox potential andror high amounts of organic matter. In case of selenium, considerable variation of distribution of species was observed. While the presence of both arsenic ŽIII. and ŽV. in the Kelheim groundwaters was obviously favored by thermodynamic parameters, this was not the case for selenium ŽIV. whose existence seemed to profit from minor concentrations of iron and manganese in the wells. Further investigations will be needed to examine the interaction of arsenic and selenium species with organic matter. There is good reason to assume that presence of humic and lignine sulfonic acids will support the existence of arsenic ŽIII. and selenium ŽIV. in groundwater samples.
Acknowledgements Michael Raessler thanks Prof K.P. Seiler, GSF National Research Centre of Environment and Health, Institute of Hydrology, for generous financial support to carry out this study. References Aggett J, Kriegman MR. Preservation of arsenic ŽIII. and arsenic ŽV. in sample sediment interstitial water. Analyst 1987;112:153᎐157. Alfthan G, Wang D, Aro A, Soveri J. The geochemistry of selenium in groundwaters in Finland. Sci Total Environ 1995;162:93᎐103. An P, Langqiu X. The effects of humic acid on the chemical and biological properties of selenium in the environment. Sci Total Environ 1987;64:89᎐98. Aurillo AC, Mason RP, Hemond HF. Speciation and fate of arsenic in three lakes of the Aberjona watershed. Environ Sci Technol 1994;28:577᎐583. Gerritse RG, Adeney JA. Rapid determination in water of chloride, sulphate, sulphite, selenite, selenate and arsenate among other inorganic and organic solutes by ion chro-
M. Raessler et al. r The Science of the Total En¨ ironment 258 (2000) 171᎐181 matography with UV detection below 195 nm. J Chromatogr A 1985;347:419᎐428. Hwang CJ, Jiang SJ. Determination of arsenic compounds in water samples by liquid chromatography-inductively coupled mass spectrometry with an in situ nebulizer-hydride generator. Anal Chim Acta 1994;289:205᎐213. Michalke B. Quality control and reference materials in speciation analysis. Fresenius J Anal Chem 1999;363:439᎐445. Mok WM, Riley JA, Wai CM. Arsenic speciation and quality of groundwater in a lead᎐zinc mine, Idaho. Water Res 1988;22:769᎐774. Raessler M, Michalke B, Schramel P, Schulte-Hostede S, Kettrup A. The capability of ultrafiltration and ETV-ICPMS for size fractionation studies of arsenic and selenium species in ground water samples with high concentrations of iron, manganese and sulfur. Fresenius J Anal Chem 1998;362:281᎐284. Raessler M. Ph.D. Thesis. Technical University Munich, Germany, 1997. Reddy KJ, Zhang Z, Blaylock MJ, Vance GF. Method for detecting selenium speciation in groundwater. Environ Sci Technol 1995;29:1754᎐1759.
181
Smith AH, Hoppenheyn-Rich C, Bates MN et al. Cancer risks from arsenic in drinking water. Environ Health Persp 1992;97:259᎐267. Tanzer D, Heumann KG. Determination of selenium species in environmental water samples using isotope dilution mass spectrometry. Anal Chem 1991;63:1984᎐1989. Van Elteren JT, Hoegee J, van der Hoek EE, Das HA, de Ligny CL, Agterdenbos J. Preservation of arsenic ŽIII. and arsenic ŽV. in some water samples. J Radioanal Nucl Chem 1991;154:343᎐347. Valentine JL, Cebrian ME, Garcia-Vargas GG et al. Daily selenium intake estimates for residents of arsenic-endemic areas. Environ Res 1994;64:1᎐9. Verordnung ueber Trinkwasser und ueber Wasser fuer Lebensmittelbetriebe. Trinkwasserverordnung 1996. German drinking water regulation, 1996. Wennrich R. Determination of arsenic species in water samples of a tin ore tailing. Vom Wasser 1997;88:1᎐12. Yokoyama T, Takahashi Y, Tarutani T. Simultaneous determination of arsenic and arsenious acids in geothermal water. Chem Geol 1993;103:103᎐111.