On-Line Coupling of Flow Injection Microcolumn Separation and

Anal. Chem. 1999, 71, 4353-4360

On-Line Coupling of Flow Injection Microcolumn Separation and Preconcentration to Electrothermal Atomic Absorption Spectrometry for Determination of (Ultra)trace Selenite and Selenate in Water Xiu-Ping Yan,*,† Michael Sperling,‡ and Bernhard Welz§

Department of Applied Research, Bodenseewerk Perkin-Elmer GmbH, D-88647 U ¨ berlingen, Germany

A flow injection manifold with an air-segmented and airtransported operational sequence for on-line coupling of microcolumn separation and preconcentration to electrothermal atomic absorption spectrometry (ETAAS) was developed for the determination of (ultra)trace selenite and selenate in water. The determination of selenite was achieved by selective reaction with pyrrolidine dithiocarbamate (PDC), sorption of the resultant Se-PDC compound onto a conical microcolumn (10.2 µL) packed with RP C18 sorbent, elution with ethanol, and detection by ETAAS. The concentration of selenate was obtained as the difference between the concentrations of selenite after and before prereduction of selenate to selenite. With the developed manifold and operation sequence, the dispersion during elution and eluate transport and the eluent volume required for complete elution of the sorbed analyte were minimized. As a result, the sorbed analyte was quantitatively eluted from the column with only 26 µL of ethanol, and all the eluate was automatically introduced into the graphite tube by an air flow without the need of preheating the graphite tube or precise timing. Pretreatment of the graphite tube with iridium as a long-term “permanent” modifier effectively prevented analyte loss arising from the high volatility of the Se-PDC compound and greatly improved the precision, sensitivity, and detection limit. One thermal pretreatment of the graphite tube with injection of 150 µg of iridium made possible at least 200 repetitive atomization cycles. With a preconcentration time of 180 s and a sample flow rate of 1.4 mL min-1, an enhancement factor of 112 was achieved in comparison with direct injection of 30 µL of aqueous solution. The detection limit (3s) was 4.5 ng L-1 Se. The RSD (n ) 7) was 3.8% at 20 ng L-1 Se. The concentrations of selenite and selenate determined in synthetic aqueous mixtures were in good agreement with the expected values. The recoveries for selenite from spiked seawater samples ranged from 98 to 102%. The concentrations of selenite in several seawater reference materials obtained with simple aqueous standard solutions for calibration agreed well with the certified and information values, respec10.1021/ac990317l CCC: $18.00 Published on Web 08/27/1999

© 1999 American Chemical Society

tively. In addition, the developed method was successfully applied to the certification of selenite and selenate in water. Electrothermal atomic absorption spectrometry (ETAAS) is a widely used sensitive technique with low detection limits for the quantification of trace elements, but direct determination of (ultra)trace elements in complex matrixes by ETAAS is usually difficult because of matrix interferences and/or insufficient detection power. Consequently, separation and preconcentration procedures, such as ion exchange, adsorption, solvent extraction, and coprecipitation, are often needed before ETAAS determination. Conventional off-line procedures for the separation and preconcentration, although effective, are usually time-consuming and tedious, require large quantities of sample and reagents, and are vulnerable to contamination and analyte loss. Flow injection (FI) on-line separation and preconcentration techniques coupled with ETAAS have been proved to be powerful for the determination of (ultra)trace levels of elements.1-5 Online sorption separation and preconcentration overcome the insufficient detection power of the analytical techniques, the matrix effects, and the shortcomings of batch-wise operation. Because complex formation and analyte extraction are often specific for only one oxidation state of an element, on-line sorption separation and preconcentration based on solvent or solid-phase extraction also offer the possibility for a differential determination of oxidation states of the element.1,6-9 Up to now, several separation techniques † Present address: Department of Geological Sciences, University of Saskatchewan, Saskatoon, SK, S7N 5E2, Canada. (Fax) (1) 306 966 8593; (email) [email protected]. ‡ (E-mail) [email protected]. § Present address: Departamento de Quimica, Universidade Federal de Santa Catarina, 88040-900 Florianopolis, S. C., Brazil. (Fax) (55) 48 331 9711; (e-mail) [email protected]. (1) Welz, B. Microchem. J. 1992, 45, 163-177. (2) Fang, Z.-L. Flow Injection Separation and Preconcentration; VCH: Weinheim, 1993. (3) Fang, Z.-L. Flow Injection Atomic Absorption Spectrometry; Wiley: Chichester, 1995. (4) Fang, Z.-L. Spectrochim. Acta, Part B 1998, 53, 1371-1379. (5) Burguera, J. L.; Burguera, M. Analyst 1998, 123, 561-569. (6) Sperling, M.; Yin, X.-F.; Welz, B. Spectrochim. Acta, Part B 1991, 46, 17891801. (7) Sperling, M.; Yin, X.-F.; Welz, B. Analyst 1992, 117, 629-635.

Analytical Chemistry, Vol. 71, No. 19, October 1, 1999 4353

have been adapted to FI on-line separation and preconcentration ETAAS systems. However, most of the work conducted dealt with preconcentration on packed columns.1-5 ETAAS is in essence not suitable for continuous processing of samples due to its discrete non-flow-through nature and the limited sample capacity of the graphite tube. To adapt an RP C18 sorbent extraction preconcentration system to the requirements of ETAAS, the eluate volume must be reduced and the dispersion must be minimized. To this end, several measures were taken before.10 A column with a volume of only 15 µL was used. The column was of conical design; the lead-diethyldithiocarbamate complex was collected on the narrow end of the column and eluted in a direction opposite to sample loading. The flow rate for elution was much lower than typically applied for flame atomic absorption spectrometry (FAAS). Even so, complete elution of the sorbed lead complex from the column still required about 200 µL of ethanol, which could not be introduced into a normal graphite tube in a single batch. To solve this problem, Fang et al.10 introduced the “eluate zone sampling” technique, exploiting the repeatability of the FI elution profile through precise timing for the collection of the fraction of the eluate bolus that contains the highest analyte concentration. However, this was found not to be rugged enough for a long-term unattended and fully automated operation.11 Additional efforts were made by Welz et al.11 to achieve a fully automatic FI on-line column separation and preconcentration procedure for ETAAS through further reduction of the column volume to 9 µL, displacement of the wash liquid with air before elution, and the use of methanol as the eluent. By these means, the analyte was eluted quantitatively with 80 µL of methanol. This volume was introduced completely into the graphite tube at a very low flow rate of 0.08 mL min-1 and the graphite tube was preheated to 80 °C.11 This quantitative transfer of the eluate, in addition to the gain in sensitivity, was considered important for a fully automatic operation. Nevertheless, the system still exhibited some problems because the eluate introduction into the preheated graphite tube had to be done very slowly and carefully and still required precise timing to avoid the entrance of unnecessary eluent into and the overflow of the eluate from the collector coil. Selenium is well-known as an essential and a toxic element to most mammalian species, including humans, depending on its chemical form and concentration.12 The very narrow concentration range for selenium between essentiality, deficiency, and toxicity12 necessitates the development of reliable analytical methods for the determination of the element in environmental and biological samples. Because the toxicity and bioavailability of selenium depend on its chemical form, it is particularly important to develop analytical methods for the selective determination of selenium compounds. For these reasons, selenium determination has been the subject of an increasing number of studies.12-26 Selenium exists (8) Yan, X.-P.; Van Mol, W.; Adams, F. Analyst 1996, 121, 1061-1067. (9) Yan, X.-P.; Kerrich, R.; Hendry, M. J. Anal. Chem. 1998, 70, 4736-4742. (10) Fang, Z.-L.; Sperling, M.; Welz, B. J. Anal. At. Spectrom. 1990, 5, 639646. (11) Welz, B.; Sperling, M.; Sun, X.-J. Fresenius J. Anal. Chem. 1993, 346, 550555. (12) Fishbein, L. In Metals and Their Compounds in the Environment. Occurrence, Analysis and Biological Relevance; Marain, E., Ed.; VCH: Weinheim, 1991. (13) Olivas, M. R.; Donard, O. F. X.; Ca´mara, C.; Quevauviller, P. Anal. Chim. Acta 1994, 286, 357-370. (14) Dauchy, X.; Potin-Gautier, M.; Astruc, A.; Astruc, M. Fresenius J. Anal. Chem. 1994, 348, 792-805.

4354 Analytical Chemistry, Vol. 71, No. 19, October 1, 1999

in several oxidation states (-II, 0, IV, and VI) in inorganic and organic forms. Selenite and selenate are present in many environmental samples, e.g., water, soils, and fly ash, and are the biogeochemically important and environmentally mobile species of the element.15-18 Most of the studies published so far on selenium speciation deal with these two species and are usually performed by hydride generation techniques or chromatographic methods with selenium-specific detectors,12-14,19-26 e.g., FAAS, ETAAS, ICPOES, and ICPMS. Because of the matrix effects and low concentrations of selenite and selenate in the environment, separation and preconcentration steps are needed before their determination.20-24 The purposes of this work were to develop an FI manifold with an air-segmented and air-transported operational sequence for direct coupling of a microcolumn separation and preconcentration technique with ETAAS to overcome the shortcomings in previous work6,7,10,11,27-29 and to apply the developed system for the determination of (ultra)trace selenite and selenate in water. EXPERIMENTAL SECTION Instrumentation. All measurements were performed on a Perkin-Elmer model 4100 ZL atomic absorption spectrometer with a transverse-heated graphite tube atomizer (THGA), longitudinal Zeeman-effect background correction, an AS-70 furnace autosampler, and a circulating cooling unit. Standard pyrolytically coated polycrystalline electrographite THGA tubes with an integrated L’vov platform (Perkin-Elmer, part no. 504033/508884) were used. A Perkin-Elmer System 2 selenium electrodeless discharge lamp was used as radiation source at the 196.0-nm wavelength with a slit width of 2.0 nm (low). The spectrometer was operated in the FIAS-furnace mode. Integrated absorbance (peak area) was used for quantitation because of its good day-to-day and tube-to-tube repeatability. The furnace temperature program for the determination of selenium is given in Table 1. The FI on-line separation and preconcentration was carried out with a Perkin-Elmer FIAS-200 flow injection accessory and an AS-90 autosampler. The analyte was collected on a PerkinElmer conical microcolumn (10.2 µL) packed with RP C18 sorbent. The standard valve of the FIAS-200 was replaced by a prototype 8-channel, 16-port multifunctional injection valve (Perkin-Elmer, (15) Masscheleyn, P. H.; Patrick, W. H., Jr. Environ. Toxicol. Chem. 1993, 12, 2235-2243. (16) Nakayama, E.; Suzuki, Y.; Fujiwara, K.; Kitano, Y. Anal. Sci. 1989, 5, 129-. (17) Tanzer, D.; Heumann, K. G. Anal. Chem. 1991, 63, 1984. (18) Niss, N. D.; Schabrob, J. F.; Brown, T. H. Environ. Sci. Technol. 1993, 27, 827-829. (19) Harrison, R. M.; Rapsomaniks, S. Environmental Analysis Using Chromatography Interfaced with Atomic Spectroscopy; Ellis Horwood Ltd.: Chichester, 1989. (20) Szpunar-Łobin´ska, J.; Witte, C.; Łobin´ski, R.; Adams, F. C. Fresenius J. Anal. Chem. 1995, 351, 351-377. (21) Larraya, A.; Cobo-Ferna´ndez, M. G.; Palacios, M. A.; Ca´mara, C. Fresenius J. Anal. Chem. 1994, 350, 667-670. (22) Pyrzyn´ska, K. Analyst 1996, 121, 77R-83R. (23) Guntinas, M. B. D.; Łobin´ski, R.; Adams, F. J. Anal. At. Spectrom. 1995, 10, 111-115. (24) Cai, Y.; Caban ˜as, M.; Ferna´ndez-Turiel, J. L.; Abalos, M.; Bayona, J. M. Anal. Chim. Acta 1995, 314, 183-192. (25) Vela, N. P.; Caruso, J. A. J. Anal. At. Spectrom. 1993, 8, 787-794. (26) Ca´mara, C.; Quevauviller, Ph.; Palacios, M. A.; Cobo, M. G.; Mun ˜oz, R. Analyst 1998, 123, 947-954. (27) Sperling, M.; Yin, X.-F.; Welz, B. J. Anal. At. Spectrom. 1991, 6, 295-300. (28) Sperling, M.; Yin, X.-F.; Welz, B. J. Anal. At. Spectrom. 1991, 6, 615-622. (29) Welz, B.; Yin, X.-F.; Sperling, M. Anal. Chim. Acta 1992, 261, 477-487.

Table 1. Graphite Furnace Temperature Program for the Determination of Selenium Using Iridium as a Permanent Modifier

Table 2. Graphite Furnace Temperature Program for Conditioning Graphite Tubes with the Iridium Permanent Modifier

time/s

time/s

step

temp/°C

ramp

hold

Ar flow/mL min-1

step

temp/°C

ramp

hold

Ar flow/mL min-1

1 (drying) 2 (pyrolysis) 3 (atomization) 4 (cleaning)

120 300 2050 2250

5 5 0a 1

30 20 5 5

250 250 0 250

1 2 3 4

110 130 1200 2000

1 20 20 1

40 50 30 5

250 250 250 250

a

Maximal power heating (∼2000 °C s-1).

U ¨ berlingen, Germany). The rotation speed of the two peristaltic pumps, their stop-and-go intervals, the actuation of the injection valve, and the interaction with the furnace autosampler arm were programmed and controlled by the FIFU version of the model 4100 ZL instrument software running on EPSON 4 × 2e PC. Ismaprene pump tubing was used to deliver the sample, reagents, and air. Small-bore (0.35-mm-i.d.) PTFE tubing was used for all connections, which were kept as short as possible to minimize the dead volume. Reagents. All reagents were of the highest available purity and at least of analytical grade. Doubly deionized water (DDW; 18 MΩ cm-1) was used throughout. A 0.05% m/v solution of ammonium pyrrolidine dithiocarbamate (APDC; Sigma) was prepared in DDW. Standard solutions of selenite and selenate were prepared fresh daily by stepwise dilution of a 1000 mg L-1 Se stock solution (Titrisol, Merck) with DDW. Nitric acid (65%, Suprapur, Merck) and hydrochloric acid (32%, Pro analysi, Merck) were used to acidify samples and standards. A 1000 mg L-1 iridium solution in 1 mol L-1 HCl (Perkin-Elmer) was used as the chemical modifier. Synthetic mixtures of selenite and selenate were prepared in DDW by mixing appropriate volumes of the standard solutions. The standard reference materials used to check the accuracy of the developed method were as follows: NASS-1, NASS-2, and NASS-4 (Open Ocean Seawater), CASS-2, and CASS-3 (Nearshore Seawater) (NRCC, Ottawa, Canada). Artificial freshwater samples spiked with selenite and selenate at about 5 and 50 µg L-1 Se, and stabilized with chloride (2000 mg L-1), were received from the Community Bureau of Reference (BCR, Brussels, Belgium) (now renamed Standards, Measurements and Testing Program) for a certification campaign.26 Procedures. The THGA tubes were pretreated with the iridium modifier three times as follows: a 50 µL aliquot of 1000 mg L-1 iridium stock solution was injected into the graphite tube and heated according to the temperature program given in Table 2. Such pretreatment was found to be effective for at least 200 atomization cycles of selenium under the temperature program given in Table 1. The developed FI manifold for the two different valve positions is shown in Figure 1. Details of the duration and function of each step are given in Table 3. A complete cycle of the separation and preconcentration without the prefill stage required 152 s with a sample loading time of 60 s. In the prefill step (Figure 1a), the column was washed with 0.03 mol L-1 HCl and the tubing was flushed with sample and APDC solution. This prefill stage was used only when a new sample was introduced but automatically

Figure 1. FI manifold for the on-line microcolumn separation and preconcentration for ETAAS: P1, P2, peristaltic pumps; MC, conical RP C18 microcolumn (10.2 µL); W, waste; EL, eluent loop; EC, eluent container; DT, delivery tube; ETA, electrothermal atomizer. Injection valve position: (a) inject; (b) fill.

omitted for replicate preconcentrations of the same sample. In step 1 (Figure 1b), the Se-PDC compound was formed on-line and sorbed onto the RP C18 column; the effluent from the column was flowing to waste. In step 2 (Figure 1a), the column was rinsed with 0.03 mol L-1 HCl solution in the counterdirection to remove residual matrix and reagent from the column. In step 3 (Figure 1a), air was introduced to remove residual acid from the column, the eluent loop (EL) and the delivery tube (DT). The ethanol drawn in this step was flowing back to the eluent container (EC) to minimize the reagent consumption. In step 4 (Figure 1b), the eluent loop was filled with ethanol, while the delivery tube was further emptied by an air flow. In step 5, with both pumps stopped, the valve was turned to the injection position and the autosampler arm automatically moved the tip of the delivery capillary into the dosing hole of the graphite tube and held near the platform. In step 6 (Figure 1a), air was drawn to propel the ethanol to elute the sorbed analyte and to transport the eluate onto the platform in the graphite tube. Finally, again with both pumps stopped, the autosampler arm automatically moved back to the wash position, Analytical Chemistry, Vol. 71, No. 19, October 1, 1999

4355

Table 3. Operational Sequence of the FI On-Line Column Separation and Preconcentration System for ETAAS step

function

valve position

duration/s

pump active

medium pumped

flow rate/mL min-1 1.4 0.9 1.4 1.4 0.9 1.4 0.9 0.5 0.8

prefill (Figure 1a)

prefill the sampling tubing

inject

20

P2

1 (Figure 1b)

load sample

fill

60

P2

2 (Figure 1a) 3 (Figure 1a) 4 (Figure 1b) 5 (Figure 1a) 6 (Figure 1a) 7 (Figure 1a)

rinse column remove residual acid fill the eluent loop capillary into the graphite tube elution and eluate introduction capillary into waste position

inject inject fill inject inject inject

20 15 6 8 35 8

P2 P1 P1

sample 0.05% m/v APDC 0.03 mol L-1 HCl sample 0.05% m/v APDC 0.03 mol L-1 HCl air ethanol

P1

air

and then the graphite furnace temperature program automatically started for the ETAAS determination. For the determination of selenate, the sample was split into two aliquots. Selenite was determined on one portion as described above, whereas the sum of selenite and selenate was determined on the other portion after prereduction of selenate to selenite by gentle boiling in 5 mol L-1 HCl medium for 50 min. The concentration of selenate was calculated as the difference. Method Development. A univariate method was used for the optimization. The integrated absorbance (peak area) was taken as the main figure of merit, but with simultaneous consideration of precision and efficiency. The fixed parameters in each study were gradually adjusted to close-to-optimal values for the final univariate studies. Three or four cycles of univariate studies were needed to establish the fixed parameters for each univariate study so that they approached optimal values. The parameters studied included sample acidity, reagent concentration, sample loading rate and time, column wash time, eluent volume, drying temperature, pyrolysis temperature, and atomization temperature. RESULTS AND DISCUSSION FI Manifold and Its Operational Sequence. Difficulties in the direct coupling of the FI microcolumn separation and preconcentration technique to ETAAS are due to the non-flow-through nature and the limited sample capacity of the electrothermal atomizer. Significant reduction in the eluent volume required for complete elution of the sorbed analyte from the column is the key to successful on-line coupling of these two techniques. To this end, an FI manifold with an air-segmented and air-transported operational sequence was developed, as described in the Experimental Section (Figure 1, Table 3). The presence of the residual acid in the column and the connecting tubing after washing the column would increase the dispersion during elution and eluate transport, and the evaporation of the resultant acid-containing eluate from the graphite tube would be much more difficult than that of the pure ethanolic eluate.11 Air was therefore introduced to remove the residual acid from the column, the eluent loop, and the delivery tube before elution (step 3) in order to minimize the dispersion and to facilitate the deposition and vaporization of the eluate. In addition, an eluent loop was used to fix the eluent volume, and air was introduced to propel the eluent and eluate, further minimizing the dispersion during elution and eluate introduction. As a result, the sorbed analyte was quantitatively eluted with only 26 µL of ethanol (see later discussion). This volume of eluate could be completely introduced into an unheated 4356 Analytical Chemistry, Vol. 71, No. 19, October 1, 1999

graphite tube at a relatively high introduction flow rate of 1.37 mL min-1, whereas in previous work11 a very low introduction rate of 0.08 mL min-1 had to be used and the graphite tube preheated to 80 °C for successful eluate deposition. Unlike previous work,6,7,10,11,27-29 it was not necessary to separate the analyte elution from the eluate introduction step in this system. The significant reduction of the eluent volume required for complete elution of the sorbed analyte and the introduction of all the eluate into the graphite tube, as well as the elution proceeding after insertion of the delivery capillary into the graphite tube, are particularly important to achieve a long-term unattended and fully automatic operation of FI on-line preconcentration ETAAS. Optimization of FI Separation and Preconcentration. The effect of HCl or HNO3 concentration on the preconcentration of selenite at the concentration of 2 µg L-1 Se was tested at an APDC concentration of 0.05% m/v. The maximal integrated absorbance was observed over an acidity range of 0.01-6 mol L-1 HCl and 0.0032-0.8 mol L-1 HNO3, respectively. The integrated absorbance under optimal conditions in the HCl medium was found to be essentially identical to that in the HNO3 medium, but the optimal acidity range was much wider in the HCl medium. Accordingly, HCl was chosen for sample acidification and a concentration of 0.1 mol L-1 HCl was used for all further experiments. Under these conditions, selenate was not preconcentrated at all. An advantage of the wide optimal range of HCl concentration was that no readjustment of the sample acidity was required for the determination of the sum of selenite and selenate after prereduction of selenate to selenite in 5 mol L-1 HCl. The effect of APDC concentration on the preconcentration of selenite was investigated in the 0.1 and 5 mol L-1 HCl medium. The results are shown in Figure 2. The maximal integrated absorbance was obtained in the range of 0.001-0.2% m/v APDC for the 0.1 mol L-1 HCl medium and of 0.02-0.2% m/v APDC for the 5 mol L-1 HCl medium. The decrease in the integrated absorbance above 0.2% m/v APDC may be caused by a competition between the Se-PDC compound and APDC for sorption at the column. The decomposition of APDC might be responsible for the slight decrease in the integrated absorbance in 5 mol L-1 HCl below 0.02% m/v APDC. For all further experiments, an APDC concentration of 0.05% m/v was used. The highest sample loading rate that could be used reliably was about 1.4 mL min-1 due to the back pressure created in a relatively thin column packed with fine-grain RP C18 sorbent. With the use of a sample loading rate of 1.4 mL min-1, the integrated

Figure 2. Effect of APDC concentration on the preconcentration of selenite at 2 µg L-1 Se with 30 s preconcentration in 0.1 and 5 mol L-1 HCl medium. All other conditions as in Tables 1 and 3.

Figure 3. Effect of the column-washing step on background (dotted line) and analyte (solid line) signals of 2 µg L-1 Se with 30-s preconcentration time: (a) in 0.1 mol L-1 HCl, without washing the column; (b) in NASS-2, without washing the column; (c) in 0.1 mol L-1 HCl, with column washing by 0.03 mol L-1 HCl in the direction opposite to sample loading for 20 s; (d) in NASS-2, with column washing by 0.03 mol L-1 HCl in the direction opposite to sample loading for 20 s. Other conditions as in Tables 1 and 3.

absorbance increased almost linearly with sample loading time up to at least 1.5 min. The influences of a column wash step on the analyte and background signals were examined with a selenite-spiked seawater sample. Without a column wash step, for aqueous standard solutions very low background and good symmetric analyte signals were observed (Figure 3a), whereas for spiked seawater, the analyte signal was heavily distorted and the background was very high (Figure 3b). This indicates that the serious interferences from seawater resulted from the matrix rather than the reagent retained on the column. Due to the selectivity of APDC, most common matrix constituents of seawater such as alkali and alkaline earth elements are separated from the analyte. However, owing to the tremendous amounts of alkali and alkaline earth metal salts in seawater, even their residues retained on the column and the connecting tubing can interfere with the subsequent ETAAS determination, producing excessively high background. Accord-

Figure 4. Effect of the eluent (ethanol) volume on the integrated absorbance of 2 µg L-1 Se in the form of selenite with 30-s preconcentration. All other conditions as in Tables 1 and 3.

ingly, a column wash step was introduced to remove residual matrix. To find out optimal conditions for washing the column, the effects of the wash medium, its acidity, and the wash time on the signals were investigated. A concentration range of 0.005-0.1 mol L-1 HCl solutions used as the wash medium was found to be optimal for maximal integrated absorbance for the analyte and minimal background. Dilute nitric acid, in contrast, was found unsuitable for the column wash because of analyte loss. A columnrinsing step with 0.03 mol L-1 HCl for 10-30 s at a flow rate of 1.4 mL min-1 in a direction opposite to sample loading was found to be very effective in removing seawater matrix without analyte loss, giving almost identical signal profiles for the analyte and background for 2 µg L-1 Se in standard solution (Figure 3c) and 2 µg L-1 Se in NASS-2 seawater (Figure 3d). Ethanol was selected as the eluent because of its effective elution of the sorbed analyte, its easy direct delivery through peristaltic pump tubing, and its relatively low toxicity. Figure 4 shows the effect of the ethanol volume on the analyte elution. The integrated absorbance for the analyte is highest between 26 and 46 µL of ethanol. This result, in addition to the fact that there is no memory effect, indicates that the sorbed analyte was quantitatively eluted from the column with 26 µL of ethanol. The decrease in the integrated absorbance of the analyte above 46 µL of ethanol probably resulted from a less favorable distribution of the analyte on the platform and/or some overflow.11 The minimal volume of ethanol required for complete elution of the sorbed analyte could not be determined in this work, because the volume of the shortest length of tubing for the eluent loop that could be mounted between the two connections was 26 µL. The much smaller eluent volume for quantitative elution in this work compared to previous work6,7,10,11,27-29 could be attributed to the significant reduction of dispersion in this air-segmented and airtransported FI system. The flow rate for the analyte elution and eluate introduction had no influence on sensitivity and precision within a range of 0.2-1.37 mL min-1. Because of the great reduction of dispersion and eluate volume, the continuous introduction of all the ethanolic Analytical Chemistry, Vol. 71, No. 19, October 1, 1999

4357

eluate and subsequent solvent evaporation were successfully performed even at a relatively high introduction rate of 1.37 mL min-1 without the need of preheating the graphite tube. Optimization of ETAAS Parameters. In this work, a preheating step was found unnecessary because only 26 µL of the ethanolic eluate was introduced into the graphite tube. The influence of the drying temperature on the determination of the analyte was evaluated with a pyrolysis temperature of 300 °C. No significant differences were observed for both the analyte and background signals within a drying temperature range of 100180 °C. Usually, a pyrolysis step is considered unnecessary for FI sorbent extraction preconcentration ETAAS because the sample introduced is only a concentrate of the analyte without any bulk materials. However, in this particular work, the first serious problem encountered was the significant loss of the analyte even at low temperatures (30 µg L-1 Cu(II), >50 µg L-1 Ni(II) and Fe(III), 100 µg L-1 As(III) and Sb(III), and >500 µg L-1 Sn(II) interfered with the preconcentration of selenite. These interferences most likely result from a competition of their ions for the complexing agent and/or of the formed complexes for the active sites of the column packing. The stability of their PDC complexes might well account for the different degree of their interferences. Performance of the FI On-Line Column Separation and Preconcentration ETAAS System. Characteristic data for the performance of the FI on-line column separation and preconcentration ETAAS system are summarized in Table 5. The proposed method was applied to the analysis of the synthetic mixtures of selenite and selenate to evaluate the possibility for a differential determination of these selenium species. The results in Table 6 demonstrate the high selectivity of the method for selenite. The determined results for selenite and the calculated values for selenate were in good agreement with the expected concentrations. To examine potential interference effects from the seawater matrix, recovery experiments were carried out by spiking seawater reference materials (NASS-1, NASS-2, CASS-2) with 0.05 and 2 µg L-1 Se, respectively, in the form of selenite. The recoveries for selenite were found to be in the range of 98-102%, indicating that the proposed method permits the determination of selenite in seawater without matrix effects. The accuracy of the proposed method was further demonstrated by determining the concentrations of selenite in seawater reference materials using simple aqueous calibration solutions. The results are given in Table 7. The determined concentrations (30) Ni, Z.-M.; He, B.; Han, H.-B. Spectrochim. Acta, Part B 1994, 49, 947-953. (31) Radziuk, B.; Thomassen, Y. J. Anal. At. Spectrom. 1992, 7, 397-403.

Table 7. Results for Selenite in Seawater Reference Materials concentration of selenite/µg L-1 Se sample

certified

determined (mean ( s, n ) 5)

CASS-3 CASS-2 NASS-4

0.020 ( 0.005

0.022 ( 0.003 0.021 ( 0.002 0.020 ( 0.005

a

0.018a

Information value provided by NRCC.

of selenite in CASS-3 and NASS-4 agreed well with the certified and information values, respectively. In addition, the concentration of selenite in CASS-2 was determined to be 0.021 ( 0.002 µg L-1 Se. Application to the Certification of Selenite and Selenate in Water. The developed method was applied to the certification of selenite and selenate in artificial freshwater samples. These samples had been spiked with selenite and selenate to about 5 and 50 µg L-1 Se, and stabilized with 2000 mg L-1 chloride. For the samples with the high spiking level, preconcentration was not necessary. Thus, the present method served as a means of separation of the analyte from the chloride matrix only and for the differential determination of selenite and selenate. To reduce the sample consumption and to maintain good precision, a sample loading time of 20 s was used, corresponding to 0.47 mL of sample solution. Five separate aliquots of each spiking level were analyzed using simple aqueous standard solutions for calibration. For the sample with the low spiking level, the concentrations of selenite and selenate were found to be 4.9 ( 0.1 and 6.7 ( 1.5 µg L-1 Se, respectively. The concentrations of selenite and selenate in the sample with the high spiking level were determined to be 34.3 ( 0.5 and 44.4 ( 1.6 µg L-1 Se, respectively. The mean concentrations of the intercalibration for selenite and selenate were 5.4 ( Analytical Chemistry, Vol. 71, No. 19, October 1, 1999

4359

1.4 and 7.7 ( 0.7 µg L-1 Se, respectively, for the sample with the low spiking level, and 35.3 ( 2.5 and 44.8 ( 4.3 µg L-1 Se, respectively, for the sample with the high spiking level.26,32 These results further demonstrate the reliability of the developed FI online column separation and preconcentration ETAAS method for differential determination of selenite and selenate in water. CONCLUSIONS This work demonstrated the feasibility of the routine determination of (ultra)trace selenite and selenate in water by on-line coupling FI column separation and preconcentration to ETAAS. With the use of the developed air-segmented and air-transported FI on-line microcolumn separation and preconcentration system, great reductions in dispersion and eluent consumption were achieved, while no precise timing was needed for elution and eluate introduction. The thermal pretreatment of the graphite tube with the iridium modifier significantly enhanced the sensitivity (32) Cobo-Ferna´ndez, M. G.; Palacios, M. A.; Ca´mara, C.; Quevauviller, Ph. Quı´m. Anal. 1995, 14 (3), 169-176.

4360 Analytical Chemistry, Vol. 71, No. 19, October 1, 1999

and greatly improved the precision and the detection limit. A single thermal pretreatment of the graphite tube with an iridium solution allows sufficient stabilization of selenium and subsequent atomization for at least 200 complete cycles. It may be expected that some elements, for which preconcentration and determination by ETAAS was not recommended because of high volatility of their dithiocarbamate complexes,33 can be determined by FI on-line sorbent extraction ETAAS, provided that graphite tubes, pretreated with a long-term “permanent” modifier, such as iridium, are used. ACKNOWLEDGMENT Presented at the 7th International Conference on Flow Injection Analysis (ICFIA95), Seattle, WA, August 13-17, 1995; Abstr. L-32. Received for review March 24, 1999. Accepted July 19, 1999. AC990317L (33) Valca´rcel, M.; Gallego, M. In Flow Injection Atomic Spectroscopy; Burguera, J. L., Ed.; Marcel Dekker: New York, 1989.