Ultratrace Determination of Inorganic Selenium without Signal

Natural surface water samples were collected and stored without headspace in 500-mL polyethylene bottles spiked with 1 mL of concentrated nitric acid...
0 downloads 0 Views 177KB Size
Anal. Chem. 2007, 79, 4558-4563

Ultratrace Determination of Inorganic Selenium without Signal Calibration Sandra G. Hazelton and David T. Pierce*

Department of Chemistry, University of North Dakota, Grand Forks, North Dakota 58202-9024

To more readily evaluate the complex biogeochemistry of selenium, a flow-through electrochemical method was developed that can accurately determine Se(IV) concentrations in aqueous samples to part-per-trillion levels without signal calibration. Stripping methods were used in conjunction with a high-efficiency, flow-through cell. The cell was designed with a novel gold working electrode that was separated from a porous counter electrode by a Nafion membrane. Because this design permitted exhaustive deposition of selenium from the sample stream as well as efficient coulometric stripping, determinations obeyed Faraday’s law over a reasonably wide range of operating conditions. The method had a minimum quantitation limit of ∼8 ng and a maximum limit of 800 ng for Se(IV). It was reliable for sample volumes as small as 0.5 mL and as high as 20 mL, thereby allowing determinations from part-per-million to just below part-per-billion levels. Interferences from Cu(II) and arsenate were evident, but only when these species were present at concentrations exceeding 10 mg‚L-1. Overall, the method had a performance comparable to hydride-generation atomic absorption spectrometry but with less cumbersome equipment and freedom from calibration. Selenium is a naturally occurring trace element that has a complex impact on the environment and living systems. As an essential nutrient in humans and livestock, it is needed at low levels for proper activity of selenoproteins and prevention of debilitating conditions such as Keshan disease in humans and white muscle disease in cattle.1 Evidence within the last several years has also shown that supplementation up to 200 µg‚day-1 in the human diet may reduce the risk of prostate, colorectal, and lung cancers.2 However, exposure of animals to high levels in the environment may be disruptive. Even though most natural waters have a total Se concentration less than 10 µg‚L-1, a variety of toxic impacts have been shown to occur through bioaccumulation.3 For sequestered ecosystems where bioaccumulation is favorable, complete reproductive failure in sensitive fish species can occur with only a few micrograms per liter of waterborne selenium.4 At higher levels, aquatic birds have experienced embryo deformity and higher adult mortality.5 * Corresponding author. E-mail: [email protected]. Phone: (701) 7772942. Fax: (701) 777-2331. (1) Combs, G. F., Jr. Br. J. Nutr. 2001, 85, 517-546. (2) Combs, G. F., Jr. J. Nutr. 2005, 135, 343-347. (3) Wu, L. Ecotoxicol. Environ. Saf. 2004, 57, 257-269.

4558 Analytical Chemistry, Vol. 79, No. 12, June 15, 2007

Because selenium can have severe impacts on living systems and the environment at trace levels or below, most physiological or biogeochemical studies have relied on the determination of Se (usually following digestion) by atomic-based spectroscopic techniques, such as hydride-generation or graphite-furnace atomic absorption spectrometry. These methods provide excellent analytical performance at the low micrograms per liter levels encountered with most samples. However, because they require cumbersome instrumentation and rigorous attention to signal calibration, they must be implemented in a dedicated laboratory setting. By contrast, it is possible with electrochemical methods to achieve these same limits of detection with easily portable equipment andsafter careful method designsfreedom from signal calibration. Stripping techniques have been the most commonly used electrochemical methods for the quantitative analysis of selenium. These include cathodic6-11 and anodic12-15 stripping voltammetry, and potentiometric stripping analysis.16-18 Several types of working electrode materials have been reported as useful in selenium determination including the following: hanging mercury drop electrodes7-10,15,19,20 and mercury film carbon electrodes;16 tin oxide;21,22 carbon23 and carbon paste;24 copper;11 electrodes modi(4) Lemly, A. D. Ecotoxicol. Environ. Saf. 2004, 59, 44-56. (5) Ohlendorf, H. M. Aquat. Toxicol. 2002, 57, 11-26. (6) Stozhko, N. Y.; Shalygina, Z. V.; Malakhova, N. A. J. Anal. Chem. 2004, 59, 374-380. (7) Sladkov, V.; Bolyos, A.; David, F. Electroanalysis 2002, 14, 128-134. (8) Mattsson, G.; Nyholm, L.; Olin, Å.; O ¨ rnemark, U. Talanta 1995, 42, 817825. (9) Quentel, F.; Elleouet, C. Electroanalysis 1999, 11, 47-51. (10) Elleouet, C.; Quentel, F.; Madec, C. Water Res. 1996, 30, 909-914. (11) Suzˇnjevic´, D.; Blagojevic´, S.; Vidic´, J.; Ercep, M.; Vucˇelic´, D. Microchem. J. 1997, 57, 255-260. (12) Bryce, D. W.; Izquierdo, A.; Luque de Castro, M. D. Anal. Chim. Acta 1995, 308, 96-101. (13) Andrews, R. W.; Johnson, D. C. Anal. Chem. 1975, 47, 294-299. (14) Cai, Q.; Khoo, S. B. Anal. Chem. 1994, 66, 4543-4550. (15) Adeloju, S. B.; Bond, A. M.; Briggs, M. H.; Hughes, H. C. Anal. Chem. 1983, 55, 2076-2082. (16) Adeloju, S. B.; Jagner, D.; Renman, L. Anal. Chim. Acta 1997, 338, 199207. (17) Gozzo, M. L.; Colacicco, L.; Calla`, C.; Barbaresi, G.; Parroni, R.; Giardina, B.; Lippa, S. Clin. Chim. Acta 1999, 285, 53-68. (18) Dugo, G.; Pera, L. L.; Pollicino, D.; Saitta, M. J. Agric. Food Chem. 2003, 51, 5598-5601. (19) Fijałek, Z.; Łozak, A.; Sarna, K. Electroanalysis 1998, 10, 846-851. (20) Korolczuk, M.; Grabarczyk, M. Electroanalysis 2003, 15, 821-826. (21) Thouin, L.; Rouquette-Sanchez, S.; Vedel, J. Electrochim. Acta 1993, 38, 2387-2394. (22) Massaccesi, S.; Sanchez, S.; Vedel, J. J. Electrochem. Soc. 1993, 140, 25402546. (23) Carbonelle, P.; Lamberts, L. J. Electroanal. Chem. 1992, 340, 53-71. (24) Carbonnelle, P.; Lamberts, L. Electrochim. Acta 1992, 37, 1321-1325. 10.1021/ac061838t CCC: $37.00

© 2007 American Chemical Society Published on Web 05/18/2007

fied with o-diamine compounds;6,25-28 and unmodified gold electrodes.12,13,15,29,30 Selenium adsorbs rather strongly to gold,13,29 making it suitable for selenium detection. The inert, nontoxic nature of gold as well as its ability to adsorb selenium without modification gives it a strong appeal as an electrode for selenium stripping analysis. Flow-through cells usually increase the sensitivity of stripping analyses by allowing large volumes of solution to pass through a relatively small-volume working electrode or working electrode compartment.31 If the electrode can be designed and operated in a manner that promotes exhaustive deposition of the analyte, it is possible to perform the stripping analysis coulometrically without the need for calibration.32-35 Charge (Q) associated with stripping the analyte from the working electrode can be calculated either by integrating the area of a stripping voltammetry peak (eq 1) or measuring the transition time (τ) of a stripping chronopotentiometry plateau (eq 2).

Q)

∫ t

0

I dt

Q ) Iτ

(1) (2)

Once Q has been determined, the absolute number of moles of analyte (N) deposited on the electrode can be calculated from Faraday’s law (eq 3, where F is the Faraday constant) provided the stripping process occurred with 100% current efficiency and the number of moles of electrons per mole of analyte (n) are known. In this situation, no calibration is needed between the measured charge and moles of analyte.

N ) Q/nF

Figure 1. Flow cell diagram. (1) gold wire working electrode (WE); (2) Nafion tubing; (3) porous stainless-steel counter electrode (CE); (4) CE electrolyte, 0.3 M Na2SO4 + 5 mM EDTA; (5) platinum wire shunt; (6) Ag/AgCl quasi-reference electrode (RE); (7) stainless steel frit; (8) reference arm electrolyte, 0.3 M Na2SO4.

(3)

To our knowledge, this paper reports the first analytical method capable of determining inorganic selenium in aqueous samples without signal calibration. The method is electrochemical in nature and relies on a novel gold working electrode to efficiently trap inorganic selenium in the selenite (Se(IV)) form. Because the method is accurate down to ultratrace levels, it may provide a reasonable alternative to traditional spectroscopic methods of analysis. EXPERIMENTAL Materials. Nitric and hydrochloric acid solutions were prepared from concentrated trace metal grade reagents (Fisher (25) Ferri, T.; Guidi, F.; Morabito, R. Electroanalysis 1994, 6, 1087-1093. (26) Yang, H.-Y.; Sun, I.-W. Electroanalysis 2000, 12, 1476-1480. (27) Yang, H.-Y.; Sun, I.-W. Anal. Chem. 2000, 72, 3476-3479. (28) Won, M.-S.; Yoon, J.-H.; Shim, Y.-B. Electroanalysis 2005, 17, 1952-1958. (29) Alanyalioglu, M.; Demir, U.; Shannon, C. J. Electroanal. Chem. 2004, 561, 21-27. (30) Gonzaga, F. B.; Pereira, C. F. Guarita´-Santos, A. J. M.; Souza, J. R. Electroanalysis 2005, 17, 2084-2089. (31) Brianina, K.; Neyman, E. Electrochemical Stripping Methods; Winefordner, J. D., Ed.; ACS Monograph in Chemical Analysis 126; American Chemical Society: Washington, DC, 1993; pp 66-67. (32) Beinrohr, E.; Cakrt, M.; Dzurov, J.; Kottas, P.; Broekaert, J. A. C. Fresenius J. Anal. Chem. 1996, 356, 253-258. (33) Sahlin, E.; Jagner, D. Electroanalysis 1998, 10, 532-535. (34) Beinrohr, E.; Cakrt, M.; Dzurov, J.; Jurica, L.; Broekaert, J. A. C. Electroanalysis 1999, 11, 1137-1144. (35) Jurica, L.; Manova, A.; Dzurov, J.; Beinrohr, E.; Broekaert, J. A. C. Fresenius J. Anal. Chem. 2000, 366, 260-266.

Scientific, Hanover Park, IL). All other reagents were analytical reagent grade and used without further purification. A primary Se(IV) stock solution (1000 mg‚L-1) was made by dissolving sodium selenite (Sigma-Aldrich, Milwaukee, WI) in 1 L of 1% HCl and was stored at 5 °C for up to 6 months. All other Se(IV) solutions were prepared from this stock on a daily basis. All solutions were prepared with Milli-Q deionized water. Flow-Through Cell. The cell (Figure 1) was adapted from a design used by Ting and Porter for electrochromatography.36 All connections were made with low-pressure Tefzel (ETFE) fittings and flangeless ferrules (Upchruch Scientific Inc., Oak Harbor, WA). The counter electrode (CE) was a length of porous stainless steel tubing (0.25-in. o.d., 0.125-in. i.d., 2-µm porosity, Mott Corp., Farmington, CT) that was fitted at each end with a standard 1/ -in. ETFE nut. The working electrode (WE) was a piece of gold 8 wire wound helically around a longer piece of straight gold wire (each 1-mm diameter, 99.9% metal basis, Alfa Aesar, Ward Hill, MA). The WE was inserted into an ethanol-swollen piece of Nafion tubing (0.108-in. o.d., 0.086-in. i.d., Perma Pure LLC, Toms River, NJ). This tubing, when dried, fit snuggly over the helical portion of the WE. This assembly was pushed inside the counter electrode, and the exposed ends of the Nafion tubing were connected at the top to an ETFE tee and at the bottom to an ETFE cross. In this configuration, the gold WE and porous stainless (36) Ting, E.-Y.; Porter, M. D. Anal. Chem. 1998, 70, 94-99.

Analytical Chemistry, Vol. 79, No. 12, June 15, 2007

4559

Figure 2. Diagram of flow cell and solution handling system.

steel CE were closely spaced but completely separated by the Nafion membrane along the entire length of the cell. The CE solution was 0.3 M sodium sulfate with 5 mM ethylenediaminetetraacetic acid (EDTA). The container for this solution was fabricated from a Teflon base and a polycarbonate tube that were sealed together as well as to the bottom nut of the CE using o-rings. A AgCl-coated Ag wire (1.0-mm diameter, annealed, 99.9%, Alfa Aesar) was used as the quasi reference electrode (QRE) in all experiments. This arm of the cell was formed using a female union and another ETFE tee containing 0.3 M sodium sulfate. The arm was isolated from the sample solution by a stainless steel frit (2-µm porosity, Upchruch). To provide better electrical performance, the QRE was connected through a 0.1 µF capacitor to a Pt wire shunt (0.5-mm diameter, 99.95% metal basis, Alfa Aesar) that was positioned as close as possible to the WE at the sample inlet. Instrumentation. Atomic absorption spectrometry (AAS) was performed with a Perkin-Elmer 560 atomic absorption spectrometer equipped with an air-acetylene burner, a burner mounting bracket holding a quartz t-tube (150-mm path length), a Se hollow cathode lamp, and a Buck Scientific 420 hydride generator. Analog output of the spectrometer was processed through an active lowpass filter (time constant 4 s) and recorded with a Hewlett-Packard model 3396 Series II integrator. Electrochemical measurements with the flow-through cell were controlled with an EG&G 273 potentiostat-galvanostat. Data were recorded with EG&G 270 Electrochemical Analysis Software v. 3.00. Numerical differentiation of chronopotentiometric data was accomplished with Origin Pro v. 7.5. Solutions were pumped through the flow cell using a multichannel Gilson Minipuls 3 peristaltic pump equipped with calibrated isoversinic tubing (Gilson Co., Inc., Lewis Center, OH). Precise sample aliquots were injected using a six-port injection valve (Upchruch) and sample loops cut from 1/8-in. Teflon tubing (0.5, 1, 2, 5, 10, and 20 mL). While sample or blank solution was flowing through the cell, two other channels of the pump were used to circulate the CE solution. A diagram of the fluid handling system is shown in Figure 2. Procedures. To ensure that no selenium remained adsorbed on the gold WE,17 it was preconditioned daily by applying alternating potentials of 0.0 V versus QRE for 10 s and +1.5 V versus QRE for 30 s while pumping a blank electrolyte of 0.1 M nitric acid at 1.0 mL‚min-1. Before analyzing each sample, the cell was flushed with 0.1 M nitric acid at a flow rate of 12.6 mL‚min-1 while applying a conditioning potential of +1.5 V versus QRE for 60 s. To begin a stripping analysis, the flow was lowered, 4560

Analytical Chemistry, Vol. 79, No. 12, June 15, 2007

a reducing deposition potential (Ed) was applied to the WE, and the sample was introduced from the six-port valve for a time needed to pass at least two loop volumes. To perform an anodic stripping analysis by chronopotentiometry (CP), blank 0.1 M nitric acid solution was pumped through the cell at 12.6 mL‚min-1 while a current of 0.0 A was applied for 20 s. After this equilibration period, the CP scan was initiated by applying an anodic stripping current (Ia). Oxidation of deposited Se(0) gave a plateau in the chronopotentiogram at ∼ +0.7 V versus QRE. The second derivative of the potentiogram was used to find the transition time (τ), which was related to the number of moles (N) of deposited selenium by using eqs 2 and 3. Dissolved oxygen had no effect on the deposition or stripping of selenium; therefore, deaeration of the sample matrix or the stripping electrolyte was not necessary. Nafion tubing was replaced weekly to maintain accurate Se determinations. The QRE maintained a potential of -0.15 ( 0.05 V versus the saturated calomel electrode (SCE) as observed on a daily basis by the anodic stripping potential of adsorbed Se(0) and the cathodic stripping of gold oxide (both at +0.8 vs SCE13). Natural surface water samples were collected and stored without headspace in 500-mL polyethylene bottles spiked with 1 mL of concentrated nitric acid. A slightly modified alkaline hydrogen peroxide digestion37 was typically used to oxidize dissolved organic matter prior to analysis. Aliquots of 1 mL of NaOH (1 M) and 2 mL of H2O2 (30%) were added to 25 mL of the filtered sample in a beaker. The beaker was covered to control spattering and was gently boiled. Aliquots of 1 mL of concentrated HCl and 3 mL of concentrated HNO3 were added to the sample, and the solution was heated in a boiling water bath for 20 min to reduce any Se(VI) to Se(IV). The final sample was cooled to room temperature and diluted to 25 mL. Determination of Se(IV) by continuous hydride-generation (HG) AAS was performed as described.38 Absorbance by atomic Se was monitored at a wavelength of 196.0 nm using a 0.7-nm band-pass and a 15-mA lamp current. RESULTS AND DISCUSSION Flow Cell Design and Performance. The cell was designed to exhaustively reduce and thereby deposit inorganic Se(IV) from an aqueous sample. Gold wire was chosen as the WE material because of its malleability and strong adsorption of Se(0). By shaping the wire into a tight helix that would fit snuggly into Nafion tubing, it was possible to minimize the ionic resistance between the WE and CE, force the sample to flow within the narrow channel of the WE helix, and prevent any mixing between the CE solution and the sample. Forcing the sample to travel within the curved channel of the WE significantly increased its contact area with the electrode and promoted efficient transport of the Se(IV) analyte to the electrode surface. Cell performance was initially evaluated by performing cyclic voltammetry of standard aqueous redox couples. For sweep rates as high as 1 V‚s-1, these couples showed nearly ideal timedependent behavior and therefore minimal iR losses or other (37) Standard methods for the Examination of Water and Wastewater, 20th ed.; Clesceri, L. S., Greenberg, A. E., Eaton, A. D. Eds.; American Public Health Association: Washington, DC, 1998; Method 3500-Se:B.3, p 3-92. (38) Standard methods for the Examination of Water and Wastewater, 20th ed.; Clesceri, L. S., Greenberg, A. E., Eaton, A. D. Eds.; American Public Health Association: Washington, DC, 1998; Method 3114C, pp 3-36-3-37.

Determination of n. The anodic stripping process for adsorbed elemental selenium (Se(0)ads) has been suggested to be a 4-electron oxidation to selenite (eq 4).13 The actual value of n was

Se(0)ads f Se(IV) + 4e-

Figure 3. Cyclic voltammograms of blank 0.1 M HNO3 (- - -) and 1 mg‚L-1 Se(IV) in 0.1 M nitric acid (s) using the designed flow cell. The WE potential was held at Ed ) -0.4 V for 2 min with a solution flow rate of 1.0 mL‚min-1. The potential sweeps were performed at 0.1 V‚s-1 (initial direction negative) without solution flowing. The labeled peaks corresponded to the following process: (C1) bulk Se(0) f Se(-II); (C2) adsorbed Se(IV) f Se(0); (A1) adsorbed Se(0) f Se(IV); (A2) bulk Se(0) f Se(IV); (A3) intermetallic [Se(0)-Au] f Se(IV).

electrical artifacts. Evaluation of solutions containing Se(IV) (as selenite, SeO32-(aq)) (Figure 3) demonstrated qualitatively similar cathodic peaks (C1 and C2) reported by Alanyalioglu et al. for Au(111) working electrode29 for the reduction of adsorbed Se(0) and Se(IV), respectively. The broad gold oxidation wave observed at ∼ +0.9 V versus QRE without selenium present, and the anodic peaks assigned to selenium (A1-A3) were all previously reported by Andrews and Johnson for polycrystalline gold electrodes.13 The oxidation peak A1 was of particular interest for performing Se(IV) determinations without signal calibration. It corresponded to the anodic stripping of Se(0) that was adsorbed to the gold surface.13 This was the only anodic stripping peak observed when fluxes of Se(IV) were low during deposition (i.e., low solution concentrations or flow rates). Peaks A2 and A3 appeared when Se(IV) fluxes were high during selenium deposition and have been ascribed to the deposition and then stripping of either bulk Se(0) (peak A2) or an intermetallic Au-Se(0) species (peak A3).13 Increases in either peaks A2 or A3 lead to proportional decreases in the stripping current for adsorbed Se(0) at peak A1. Both stripping voltammetry and chronopotentiometry were evaluated for determination of Se(IV). Chronopotentiometry (CP) plateaus yielded more reproducible charge measurements than stripping voltammetry peak areas and were used for subsequent determinations. The plateau corresponding to anodic stripping of adsorbed Se(0) occurred at ∼ +0.7 V versus QRE (Figure 4A, curve 5). Although a stripping plateau was readily apparent for nanomole quantities of adsorbed Se(0), it was difficult to precisely measure τ for subnanomole quantities (Figure 4A, curves 2 and 3). This problem was overcome by numerically calculating the second derivative of each chronopotentiogram39 and measuring the time between the indicated minimum and maximum (Figure 4B). By using this technique, it was possible to reliably determine quantities of adsorbed Se(0) down to 0.1 nmol (∼8 ng of Se(IV)). (39) Guenter, H. Fresenius J. Anal. Chem. 1983, 315, 438-447.

(4)

determined by depositing Se(0)ads from sample aliquots containing varied quantities of Se(IV) and measuring the anodic stripping charge (Qobs). The number of moles of Se(IV) (N) was varied by using different concentrations of Se(IV) (2-1000 µg‚L-1) with the same sample loop volume of 5 mL and by using different loop volumes (1-20 mL) of the same 20 µg‚L-1 solution of Se(IV). If exhaustive deposition of Se(0)ads was achieved and the anodic stripping process was 100% efficient, a Faraday’s law plot of Qobs/F against N would be linear with a slope equal the apparent n-value. The data (Figure 5) showed a distinctly linear region below 10 nmol of Se(IV), which corresponded to an n-value of 4.03 ( 0.07. An expanded graph of this region has been provided as Supporting Information (Figure S1). Above 10 nmol (∼800 ng of Se(IV)), the stripping of Se(0)ads showed a negative deviation from Faraday’s law. Several effects may have contributed to this deviation. Nonexhaustive deposition was the most likely effect, and it was detected by HG-AAS of the cell effluent. However, this breakthrough of Se(IV) was only detected at the highest flux condition (1000 µg‚L-1), and the amount of Se(IV) recovered was far too small to complete the mass balance. Formations of bulk Se(0) and intermetallic Au-Se(0) species under high-flux conditions were also possible because cyclic voltammetry showed these processes competed with the deposition of Se(0)ads. However, separate CP stripping plateaus were not discernible for either of these competing Se(0) forms, and it was uncertain whether they played a significant role in the mass balance. These results confirmed that the cell was able to exhaustively reduce Se(IV) to Se(0)ads, at least under low-flux conditions, and that anodic stripping of this layer occurred with near 100% efficiency by a 4-electron process; most likely the process described in eq 4. These characteristics made it possible to determine Se(IV) using Faraday’s law (eq 3) over a reasonably wide range of conditions and effectively eliminated the need for signal calibration. Because the method was reliable for sample volumes as small as 0.5 mL and as high as 20 mL, determinations could be performed for Se(IV) concentrations as high as partsper-million to just below parts-per-billion. Optimizations. A number of operating parameters affected both the accuracy and precision of the flow cell method and required optimization. Optimization graphs for determination of 1.27 nmol of Se(IV) have been provided as Supporting Information (Figures S2-S5). Based on these data, optimum flow rates, deposition potential, and stripping current as well as their acceptable operating ranges were determined (Table 1). Matrix Interferences. Because a flow-through cell was used, the sample matrix was only exposed to the working electrode during the deposition step and was subsequently flushed out of the cell. This design allowed the introduction of a reproducible electrolyte (0.1 M nitric acid) during the stripping step and eliminated one avenue for interference. Nevertheless, certain inorganic ions and acids in the sample matrix did affect the Analytical Chemistry, Vol. 79, No. 12, June 15, 2007

4561

Figure 4. Anodic stripping potentiograms (A) and their second derivatives (B) following cathodic deposition from 2.0-mL standard samples with (1) 0.0, (2) 0.253, (3) 0.506, (4) 1.27, and (5) 2.53 nmol of Se(IV). Potential axis in graph A corresponds to curve 5 (note arrow). Curves 1-4 in both graphs have been offset vertically for clarity. Transition time boundaries have been marked (+). Samples were analyzed using Ia ) 3.0 mA, Ed ) -0.5 V, deposition flow rate 1.0 mL‚min-1, and stripping flow rate 12.6 mL‚min-1. Table 1. Optimized Method Parameters for Calibration-Free Selenium Determinations Ed (V vs QRE)

Ia (mA)

optimum value -0.5 3.0 operable range -0.7a to -0.4 2.5-5.0 tested range -0.9 to 0.2 0.5-5.0b

deposition stripping flow rate flow rate -1 (mL‚min ) (mL‚min-1) 1.0 0.5-1.0 0.5-12.6c

12.6 5.0-12.6 0-12.6c

a Deposition potentials below -0.7 V caused significant dissolution of the stainless steel counter electrode. b Stripping currents above 5 mA decreased precision because of very rapid transitions and the finite data acquisition rate of the potentiostat/galvanostat. c Maximum flow rate for the pump tubing used.

Figure 5. Relationship between the moles of Se(IV) (N) that passed through the flow cell during cathodic deposition and the moles of electrons (Qobs/F) consumed by anodic stripping of adsorbed Se(0). Filled circles (b) represent different concentrations of Se(IV) (2-1000 µg‚L-1) with the same sample loop volume of 5 mL, and open circles (O) represent different loop volumes (2-20 mL) of the same 20 µg‚L-1 Se(IV) solution. Error bars reflect ( one standard deviation of triplicate measurements. Samples were analyzed using Ia ) 3.0 mA, Ed ) -0.5 V, deposition flow rate 1.0 mL‚min-1, and stripping flow rate 12.6 mL‚min-1.

deposition of Se(0)ads or the ability to distinguish the CP stripping plateau for Se(0)ads. Strong inorganic acids are often added to selenium samples in order to perform some level of matrix digestion prior to analysis. Hydrochloric acid is particularly common because of its use in reducing Se(VI) (selenate) to Se(IV) (selenite).40 However, chloride ions adsorb to gold and can potentially interfere with anodic stripping of Se(0)ads. Flushing the WE with 0.1 M HNO3 prior to stripping was sufficient to remove most adsorbed chloride. The very small amount that remained actually benefited the stripping analysis by shifting the gold oxidation process to more positive potentials and increasing the resolution of the CP stripping plateau for Se(0)ads. Although high concentrations of other strong acids seemed to prevent the deposition of Se(0)ads, these effects could be eliminated by diluting the samples until the acid was 1 M or less. Numerous metal ions were tested as potential matrix interferences. All of these tests were conducted with a Se(IV) concentra-

tion of 10 µg‚L-1 and a 2-mL sample loop. Only Cu(II) and arsenate were found to interfere, but at concentrations above 10 mg‚L-1. These interferences were caused by anodic stripping processes associated with Cu(II) and arsenate, which competed with stripping process for Se(0)ads. Other species investigated for interference were phosphate, Fe(III), Zn(II), Mg(II), Mn(II), Mo(II), Al(III), Cr(III), Co(II), Ni(II), selenate, and Cd(II). Phosphate did not interfere up to 0.1 M, and the other metal ions did not interfere up to the tested limit of 10 mg‚L-1. Notably cadmium did not show an effect, although it has been reported to interfere with Se(IV) determination at a gold electrode with a mole ratio of only 0.5:1.12 Natural matrixes were also evaluated for matrix interferences because dissolved organic matter is known to affect the determination of Se(IV) by HG-AAS, possibly through direct interactions with the selenite ion.41 To gauge whether a similar effect occurred with the flow cell, natural and organic-free samples were tested for Se(IV) spike recovery. Untreated, both deionized and bottled water samples demonstrated recoveries that were within the 95% confidence limit of the actual amount of Se(IV) added. However, the recoveries for untreated natural water were consistently high and quite variable (Table 2). Oxidative digestion significantly improved the recovery of Se(IV) from the natural matrix, suggesting that dissolved organic matter contributed to the interference. However, one drawback

(40) Bye, R. Talanta 1983, 30, 993-996.

(41) Roden, D. R.; Tallman, D. E. Anal. Chem. 1982, 54, 307-309.

4562 Analytical Chemistry, Vol. 79, No. 12, June 15, 2007

Table 2. Spike Recoveriesa from Water Samples Determined by the Flow Cell Methodb

samplec deionized water deionized water deionized water deionized water bottled water I bottled water I EC surface water EC surface water

total Se in original Se(IV Se(VI) sampled added added % recovery treatment (µg‚L-1) (µg) (µg) of total Se none digestion none digestion none digestion none digestion