Beyond the Hydrogen Wave: New Frontier in the ... - ACS Publications

ARTICLE pubs.acs.org/ac

Beyond the Hydrogen Wave: New Frontier in the Detection of Trace Elements by Stripping Voltammetry Pascal Sala€un,* Kristoff Gibbon-Walsh, and Constant M.G. van den Berg Oceans and Ecosystems, School of Environmental Sciences, University of Liverpool, 4 Brownlow Street, Liverpool, United Kingdom, L69 3GP

bS Supporting Information ABSTRACT: Stripping voltammetry is limited in acidic conditions to relatively high deposition potentials because of the interfering effects of the hydrogen produced at the working electrode. We report here a simple procedure to perform reliable and sensitive trace metal analysis in such conditions. Measurements are made at a gold microwire electrode. After applying a simple electrochemical conditioning procedure, hydrogen does not block the electrode, allowing reproducible analysis and smooth stripping signals to be obtained. Advantages of working inside the hydrogen wave are exemplified through the detection of the often considered electroinactive antimony(V). Sb(V) is detected in relatively low acidic conditions (pH e 1) using low deposition potentials (e
1.8 V). The detection limit is 5 pM (0.63 ppt), the lowest ever reported for an electroanalytical technique and one of the lowest analytical methods. The method is simple, robust, and free from the common arsenic interference due to selective electrochemical hydride generation of arsine over stibine during the deposition step. Analytical methods were optimized and tested on mineral, river, tap, and coastal seawater. Results favorably compare against Certified Reference Materials data (NASS-4 and SLRS-3) and ICPMS analysis. Deposition well below the hydrogen wave pushes the frontier of stripping voltammetry, and new analytical applications in this combined range of acidity and deposition potential are to be expected.

A

nodic stripping voltammetry is a powerful analytical technique for trace metal detection. It first consists of an electrolysis/deposition step where the metal of interest is accumulated at the surface of the working electrode (WE) during a certain time. This is followed by a stripping step which removes the previously accumulated metal back into the solution, the current generated during the stripping being directly proportional to the concentration of the metal present in the water sample. During the electrolysis step, the metal is reduced at the WE together with any other elements that can be reduced at this deposition potential. If a low deposition potential is used in acidic conditions, hydrogen is formed at the WE surface by the reduction of protons and oxidants are produced at the auxiliary electrode. The hydrogen generation is problematic in many ways: it blocks the electrode surface, affects the reproducibility, and increases the noise of the voltammograms. The very few attempts that have been made to measure in such conditions were mostly directed toward the determination of As(V) which is only reducible in acidic conditions. To avoid hydrogen bubbles blocking the electrode, a rotating electrode was developed with the gold disk placed on the side to facilitate the removal of hydrogen.1 Measurements were made in 0.25 M acid using a deposition potential of
1.1 V, but the gold disk needs to be polished and cleaned after only 10 measurements. The wire electrode has been more successful. Using a gold film deposited on a Pt wire electrode, As(V) was detected in 0.2 M HNO3 with Edep=
1.8 V in 0.2 M2 and at
1.6 V in 0.1 M HCl.3 In these two r 2011 American Chemical Society

cases, the gold film needs to be replaced regularly and both methods used a flow system to change the solution between the electrolysis and the stripping steps (medium exchange), which slightly complicates the analytical system although it offers rapid cleanup and high sample throughput. Recently, we reported a simpler procedure for As(V) analysis in 0.1 M HCl using Edep =
1 V at a solid Au polycrystalline wire electrode4 where no medium exchange nor film deposition were required. In this case, although hydrogen is produced during the deposition step, the stripping signal is smooth and very stable, showing a negligible effect of the hydrogen produced during the deposition. In this work, we have assessed the potential of the gold wire electrode for measurements in more drastic conditions of acidity and deposition potential, well into the hydrogen wave, i.e., in a region where stripping analysis is always avoided. To best exemplify the capability of the gold wire to perform in such conditions, we chose antimony as our target because of its perceived electro-inactivity and because of its growing environmental importance. Antimony is a nonessential element and suspected human carcinogen5 and, as such, is one of the “high priority” pollutants defined by the EU.6 It is released in the environment mainly Received: February 6, 2011 Accepted: April 2, 2011 Published: April 02, 2011 3848

dx.doi.org/10.1021/ac200314q | Anal. Chem. 2011, 83, 3848–3856

Analytical Chemistry through anthropogenic activities (flame retardants, brake pads, plastics), and growing concerns emerged when polyethylene terephthalate (PET) bottles were found to leach nM levels in the mineral water.7 Typical levels in aqueous systems are low with concentrations around 1.5
2 nM (∼0.2 ppb) in seawater8 and less than 8 nM (∼1 ppb) in rivers and lakes8 but can be as low as a few tens of pM (ng/L) in pristine groundwaters.9 Antimony is a metalloid and forms strongly hydrolyzed inorganic species (e.g., Sb(OH)3 and Sb(OH)6
) in neutral aqueous solutions. In oxygenated water, it is mostly present as antimonate (Sb(V)) and to a lesser extent antimonite (Sb(III)); monomethyl MMSb and dimethyl DMSb stibonic acids can also be present but at usually much lower concentrations.8 Electrochemical methods for the determination of total antimony Sb have mostly focused on the detection of Sb(III) because of the perceived electroinactivity of Sb(V). Prior to detection, the sample must undergo a chemical reduction step to transform all Sb(V) into Sb(III) using, e.g., L-cysteine,10 sulfur dioxide,11 hydrazine dihydrochloride,12 or potassium iodide.13 This reduction step is time-consuming, might be difficult to implement, and is a major potential source of contamination at low Sb levels. Although Sb(V) is often considered to be electroinactive, its electrochemical reduction has been observed in high acidic chloride containing solution,14 which is attributed to the formation of hydroxyl-chloro complexes (e.g., Sb(OH)3Cl3
) that are more easily reduced than the corresponding strongly hydrolyzed Sb(OH)6
. However, this direct reduction has only been used in very few cases: in up to 5 M HCl on Hg electrodes15 and, more recently, on a tubular gold electrode in 5.5 M HCl.16 The lowest limit of detection for Sb(V) direct determination by ASV is 0.27 nM using 10 min deposition.15b Although these methods do not require a chemical reduction step, the high acidic conditions are not appealing for routine use and are again source of potential contamination. To date, the most sensitive electroanalytical method is by cathodic stripping voltammetry (CSV) on in situ deposited bismuth film in mild acidic conditions (0.6 M HCl).17 Detection limits are in the low ppt range (16 pM). Bi(III) is deposited together with Sb(V) and promotes the reduction of Sb0 to stibine (SbH3) in the cathodic stripping scan. This work presents an alternative electroanalytical method for the detection of antimony at trace levels in natural waters including seawater using a mercury-free, solid, electrode. The method is relatively simple but very sensitive, can be performed in mild acidic conditions (pH 1), and can easily be adapted for field monitoring. In addition, this work lays out experimental conditions to perform stripping analysis with deposition potential well below the hydrogen wave.

’ EXPERIMENTAL SECTION Chemicals. Water used to prepare reagents and preliminary working solutions was deionized water from a Millipore system (resistivity 18 MΩ.cm). HCl was purified by double sub-boiling distillation on a quartz condenser. Standard Sb(V) and Sb(III) solutions (10
4 to 10
8 M) were prepared from 1000 ppm standard solution from Aldrich and BDH, respectively. They were acidified to pH 2 (HCl) and kept at room temperature. Standard solutions were found stable for at least 6 months. Flasks containing Sb(III) standards solutions were wrapped in aluminum foil paper to avoid oxidation to Sb(V). HNO3, HClO4, and H2SO4 were from BDH Chemicals Ltd. (AnalaR grade, England). Metal solutions (Cu, Pb, Hg, Mn, Zn, Cd, Ni, Fe, Bi(III),

ARTICLE

and Se(IV)) used for interference experiments were diluted from AAS standard solutions from Fluka, Aldrich, or VWR; sodium dodecyl sulfate (SDS) was from Aldrich, and Triton X-100 was from BDH (England). Fulvic acid (FA) and humic acid (HA) standards were provided by the International Humic Substance Society (IHSS). The source of these standards is the Suwannee River. A second humic standard from Aldrich was also used. Equipment, Electrode Fabrication. Voltammetric experiments were made using a computer controlled μAutolabIII potentiostat (Eco Chemie, Netherlands) through the associated GPES software (version 4.9). A three electrode glass cell containing the working electrode (WE), an iridium counter electrode (CE; 150 μm diameter, ∼3 mm length), and a double-junction, Ag/AgCl/KCl (3 M)//NaNO3 (0.1 M), reference electrode was placed in a voltammetric cell in a Faraday cage (Windsor Scientific, UK). The working electrode was used either with a standard magnetic stirrer (728 Stirrer
Metrohm) or with a vibrating device such as described previously18 (1.5 V, 200 Hz frequency). The potentiostat was a μAutolabIII connected to the appropriate interface (IME663-Autolab) and controlled by the GPES software (Version 4.9). The gold microwire electrodes were prepared as described previously4,19 with minor changes: a copper wire (100 μm diameter, ∼10 cm long) is first placed through a pipette tip, and its end is dipped into a conductive solution (Leitsilber L100), which has excellent adhesive properties. An electrical connection was made by touching the gold wire, and the copper wire was then pulled back until the gold wire reached the pipette tip. The tip was melted around the base of the gold wire either with a heat gun or in a specifically designed oven set at 450 °C. The tip was connected to an electrical wire covered with conductive resin (SL65-RiteLok) to ensure a good contact with the copper wire. No hardener (SL65-Rite Lok) was used as it was suspected to create electrical noise. The electrodes were stored in air or Milli-Q water. Contrary to the disk, the gold wire did not need polishing. The surface of the electrode was cleaned electrochemically by hydrogen generation at e.g.
3 V for 60 s, after which its behavior was checked in 0.5 M H2SO4 by running cyclic voltammograms between 0 and 1.5 V.19 Electrode Real/Geometric Area and Diffusion Layers. The real surface area was calculated from the charge corresponding to the reduction of one oxide monolayer in 0.5 M H2SO4 as previously described.19 The geometric area was obtained by fitting the chronoamperommetric current obtained in a stagnant solution of 0.5 M KCl þ 10 mM K3FeCN6
to the theoretical, predicted, current. The size of the diffusion layer was calculated from the chronoamperometric current obtained under vibrated conditions.19 Voltammetric Determination of Sb(V) in Natural Waters. Sb(V) in natural waters was determined by anodic stripping voltammetry (ASV). Both square wave (SW) and differential pulse (DP) were used, although DP was found to display a better signal/noise ratio resulting in lower detection limit. In seawater, background subtraction was used. It consists in the successive recording of two scans: the analytical scan followed by a background scan (same as the analytical scan but with only few seconds of deposition). The background is subtracted from the analytical to give a background corrected scan, which was used for quantification. In freshwater and tap and mineral water, background subtraction is not required, as the baseline is relatively flat. Both peak heights and/or derivative were determined automatically with the GPES software, even at low Sb levels. The entire analytical procedure is automated 3849

dx.doi.org/10.1021/ac200314q |Anal. Chem. 2011, 83, 3848–3856

Analytical Chemistry

ARTICLE

Figure 1. Stripping analysis with deposition beyond the hydrogen wave. (A) Voltammograms obtained at Edep = 0.24,
0.6,
1.2,
1.8,
2.4, and
3 V. (B) Relative intensity as a function of deposition potential (normalized at
3 V). Solution: 0.5 M H2SO4 þ 0.1 M HCl þ 5 nM Sb(V), 10 nM Cu and 10 nM Hg; Electrode: 5 μm gold, Edep (60s),
1.5 V (1s), 1s equilibrium, stripping from
0.1 to 0.55 V (DP, 4 ms modulation time, 100 ms interval time, 4 mV step, 50 mV amplitude).

through the Project option of the GPES software. All concentrations determined by standard addition consisted of a minimum of three additions and three measurements per addition with automatic peak search. Original concentrations and associated standard errors were calculated in Excel (LINEST function) using all measurements (N g 12).

’ RESULTS AND DISCUSSION Optimum Experimental Conditions. Electrode Conditioning. The surface of the electrode was activated using strong

hydrogen evolution in 0.5 M H2SO4 by imposing a potential (Eclean) of
3 V (for the 5 μm diameter electrode) and of
4 V (for electrodes of 10 or 25 μm diameter) for typically 30
120 s. It was found that, without this cleaning step, deposition within the hydrogen wave creates a buildup of hydrogen bubbles at the electrode surface, held in place by the electrode body. The voltammetric signal obtained during the stripping scan was then perturbed (noise, irreproducibility) in a similar way as what can be observed at a macro-disk electrode. However, after the cleaning step, strings of small H2 bubbles can be seen to stream off the WE into the solution without any buildup at the electrode surface. The stripping signal is then smooth, and the analysis is reproducible. Repeated imposition of
5 V for any length of time (up to 5 min) in acid up to 1 M H2SO4 did not modify the surface area (the gold oxide reduction peak in CV remained the same). An increase of the roughness was therefore not the reason for the easy removal of hydrogen but probably the removal of an adsorbed organic layer at the surface of the gold, which may have been deposited on the gold during the fabrication process or during storage in the plastic container. This is consistent with alternative, successful, conditioning treatments also involving oxidation of the gold surface through UV irradiation or by placing the electrodes (gold and copper wire embedded in the pipet tip) in a boiling oxidizing solution (30 min in 0.5 M H2SO4 þ 10 mM persulfate). These may be useful when a potentiostat is used with a potential range limited to (2 V or with a limited maximum current (e.g., 0.5). Measurement Stability. The stability of the peak for Sb(V) was found to be affected by high deposition currents. The peak intensity of repeated scans decreased with time when deposition took place at high current (e.g., pH 0, Edep =
2.5 V), while it was stable at relatively low deposition current (e.g., pH 1, Edep=
1.8 V; Figure S-3, Supporting Information). This effect could be attributed to increasing amounts of oxidant produced at the auxiliary electrode (chlorine, persulfate, or products of nitrate oxidation from HCl, H2SO4, and HNO3 solutions, respectively). A compromise needs to be found between stable analysis (relatively low acidic conditions and high deposition potential) and high sensitivity (relatively high acidic conditions and low deposition potential). Any variables that affect the concentration of these oxidants (solution volume, working and auxiliary electrode length, deposition time, deposition potential, acidity) may influence the stability. It is important to note that this loss of stability is metal dependent with Cu being unaffected by the experimental conditions (Figure S-3, Supporting Information). The loss of Sb signal is therefore due to chemical reasons and not to a degradation of the electrode. Influence of Electrolyte. The stripping voltammetric response of 20 nM Sb(V) was first assessed by varying the deposition potential in HCl solutions of different strengths (Figure 4A). No signal was obtained when pH > 1 or ifEdep >
0.8 V. From pH 1 and below, the Sb(V) signal increased with decreasing deposition potential and pH. The response stabilized at HCl concentration >1 M, with no further increase up to 4 M HCl (Edep =
2 V). Similar behavior was obtained for perchloric, nitric, and sulphuric acids (Figure S-4, Supporting Information) between pH 2 and 0, with a leveling of the signal for [Hþ] ∼ 1 M. Peak intensity and potential for Sb was similar (within 80 mV) in HNO3, HClO4, and H2SO4. Optimum sensitivity was obtained in HCl, where the peak height was greater (slightly thinner peak 3852

dx.doi.org/10.1021/ac200314q |Anal. Chem. 2011, 83, 3848–3856

Analytical Chemistry

ARTICLE

Figure 4. Influence of the deposition potential on (A) Sb(V) and (B) As(V) peak intensities in HCl solutions of different strengths. Error bars represent the standard deviation of three consecutive measurements.

with W1/2 = 32 mV for HCl = 1 M) and the peak potential slightly more negative (20 mV) which might indicate the formation of Sb(OH)xCly complexes. In HNO3, HCl, and H2SO4, a shoulder occurred on the positive side of the stripping peak, whereas in HClO4, an additional peak occurred at ∼200 mV. The reason for the shoulder and additional peak was not investigated, but it is likely that they result from the oxidation of stripped Sb(III) into Sb(V), similar to arsenic which showed a secondary peak due to oxidation of As(III) to As(V).24 Interferences. Arsenate, Metals, and Surfactants. In 0.1 M HCl, the As(V) peak is only 50 to 70 mV more negative than the Sb(V) peak.4 Natural levels of arsenic is usually an order of magnitude higher than those of Sb; a reliable determination of Sb is difficult if the two peaks overlap. In an attempt to separate them, the influence of the deposition potential on their respective peak intensities was studied at various HCl concentrations (Figure 4). At each HCl concentration, the As(V) signal initially increased with decreasing Edep, reached a maximum (Edep ∼
1.1 V), and then decreased until complete loss (Figure 4B). This decrease is attributed to the formation of arsine (AsH3) during the deposition step through reduction of As(V) to adsorbed As° followed by its multistep reduction20a to AsH3. Such reduction of As(V) to AsH3 in acidic media has previously been observed at materials with both high and low overvoltage for H2 generation, on Hg25 and on vitreous carbon,26 and was also found on platinum.20a By comparing Figure 4A,B, one can see that As(V) is only measured in conditions of low acidity (pH 1
2) using a relatively high deposition potential (
1 V) while Sb(V) is detected without As at high acidity (pH < 1) and low deposition potential (Edep <
1.8 V). The same was found in seawater: addition of 50 nM Sb(V) did not affect the As peak (pH 1, Edep =
1 V) and addition of 500 nM As(V) did not affect the Sb peak (pH 1, Edep =
1.8 V). Possible interference of Sb(III), As(V), As(III), MMA and DMA (monomethyl and dimethyl arsenous acid), Pb, Zn, Cd, Bi, Mn, Cu, Se, Ni, Triton X100, and SDS were evaluated on the Sb peak obtained in acidified seawater (addition of 0.1 M H2SO4) at environmentally relevant and greater concentrations. Voltammetric deposition was for 60 s at
1.8 V. The same sensitivities were obtained for Sb(III) or Sb(V) due to the fast oxidation of Sb(III) to Sb(V) with the chlorine which had been electrochemically produced at the auxiliary electrode during the deposition step, similar to

the oxidation of As(III) to As(V).4 Therefore, in these measuring conditions, all inorganic Sb was present as Sb(V). Additions of 20 nM MMA, 20 nM DMA, 10 nM Pb, 10 nM Cd, 100 nM Zn, 100 nM Se, or 500 nM Mn(II) did not interfere. Interfering effects were caused by addition of Ni, Bi, and Cu: addition of 40 nM Ni resulted in a 45% decrease, and addition of 50 nM Cu and 1 nM Bi resulted in an almost complete loss of the Sb peak signal. Natural Bi levels in seawater are much lower than nM levels and do not interfere, but for higher levels, interference can be prevented by addition of iodide.27 The Cu interference can be minimized using a high acidity and low deposition time to specifically obtain the Sb signal. However, for Cu concentration less than 25 nM, no significant interference was observed. The interference is characteristic of the formation of a intermetallic compound between Cu and Sb. These compounds are known to occur at a Hg film electrode11a and might thus also occur on the gold solid electrode, although the chemical state of the reduced species is fundamentally different on the gold than in Hg (soluble). However, even in tap water where Cu concentration was very high (∼4 μM), the antimony produced a linear response up to 500 nM and the Cu peak remained constant (Figure S-5, Supporting Information), thus excluding the formation of interfering Cu
Sb intermetallic compounds in this matrix. In addition, the resolution between Cu and Sb was improved by increasing the acidity, decreasing the deposition time, and increasing the chloride concentration (Figure S-5, Supporting Information). While the Cu peak potential was almost insensitive to the chloride concentration in this range (0.1 to 5 M), the Sb peak potential was moved more negative with increasing chloride which improved the peak separation. In 3 M HCl, using Edep =
1.8 V for 2 s (stagnant conditions), the detection limit was 0.96 nM with square wave as stripping procedure (50 Hz, 25 mV, 6 mV). Original concentration in this tap water was below this detection limit. Addition of 3 ppm SDS resulted in a 15% decrease, but only a 30% decrease was observed for 30 ppm. Addition of 1 ppm Triton X100 did not affect the peak while a 20% decrease was observed for 2 ppm additions and a 75% decrease was obtained for 5 ppm. This is in contrast with the analysis of arsenate where 30 ppm SDS and 5 ppm Triton did not affect the peak.4 Organics. Humics acids (HA, Suwanee River and Aldrich) and fulvic acids (FA, Suwanee River) were used as terrestrial model compounds. Alginic acid (AA) is a polysaccharide present in the 3853

dx.doi.org/10.1021/ac200314q |Anal. Chem. 2011, 83, 3848–3856

Analytical Chemistry

ARTICLE

Table 1. Comparison of Sb Values Determined by Voltammetry (DP, 4 ms Modulation Time, 100 ms Interval Time, 50 mV, 6 mV)a

sample

added electrolyte

background subtraction

Edep (V) tdep (s)
1.8 60
1.8 60
2 8
2 13
2 13
2 18
2 13
1.8 300

NASS-4 (coastal water)

0.1 M HCl

yes

NASS-5 (coastal water)

0.1 M HCl

yes

SLRS-5 (river water)

1 M HCl

no

Highland Spring still (Scotland)

0.2 M H2SO4 þ 10 mM HCl

no

Evian (France)

0.1 M H2SO4 þ 0.1 M HCl

no

Buxton water still (England)

same

no

Highland spring sparkling (Scotland)

0.2 M H2SO4 þ 30 mM HCl

no

Milli-Q water

0.2 M HCl

no

sensitivity (pA/ V 3 nM 3 s 3 nC)

total Sb found (nM)

total Sb certified (nM)

% difference

49.5

1.50 ( 0.10

16.7 ( 6.1

65.8

0.99 ( 0.06

1.80 ( 0.16a 1.72b no values

n.a.

110.0

2.82 ( 0.12

2.46b

14.6

150.0

1.62 ( 0.07

1.53c

5.9

101.7

2.38 ( 0.10

2.55c

6.7

93.6

1.69 ( 0.09

1.70c

0.6

82.1

1.80 ( 0.07 2.01 ( 0.15 bdld

2.67c

32.6 24.7 n.a.

96.44

n.a.

a Literature (see ref 31). b Information value given by CNRC (Conseil National de Recherches Canada). c Values obtained by ICPMS. d Below detection limit.

cell walls of brown algae and was used as exudate model compounds. HA, FA, and AA were added to UV digested seawater acidified with 0.1 M H2SO4. Measurements were made at natural concentration levels immediately after additions. A 10
15% decrease was observed for concentrations as low as 0.5 ppm of HA and AA while a 15% decrease was observed at 3.5 ppm of FA. The absence of peak potential shift and change in peak half width suggest the formation of electrochemically inert complexes that have been reported to occur in acidic conditions in sediments,28 freshwater,28,29 and coastal seawater.11b Figures of Merit. The electrode was chosen small (5 μm diameter, short length l ∼ 200 μm) to limit the amount of oxidants produced at the counter electrode. Clean triple distilled HCl acid was used. An acid clean Teflon cell was used to limit Sb adsorption on the walls, which was found to occur when a glass or a quartz cell was used. At these low Sb levels, great care should be taken to avoid contaminations and potential adsorption problems.9 Measurements were done here in 0.2 M HCl with Edep =
2.0 V using DP as stripping method (100 ms interval, 4 ms modulation time, 50 mV amplitude, 4 mV step). Using tdep = 60 s, the response was linear at least up to 100 nM and the relative standard deviation of 40 analysis was less than 5%. Using tdep = 300 s, the response was found linear (r2 = 0.9984) between 0 and 250 pM (50 pM additions, Figure S-6, Supporting Information) highlighting the potential of the technique to measure reliably low Sb levels. The detection limit (DL) was calculated at 5.4 pM or 0.63 ppt (3 times the standard deviations of 7 consecutive analysis at 20 pM added Sb). Similar DL (5.6 pM) could also be obtained with a 10 μm electrode or using 0.1 M HCl (5.9 pM). To our knowledge, these are the lowest DLs ever reported for an electroanalytical method and are at least ∼100 times lower than any previously reported method with direct reduction of Sb(V). It also compares favorably with most spectroscopic and hydride techniques30 while being a relatively cheap and simple method to implement. Such DLs should allow the analysis of pristine groundwaters9 if care is taken to avoid contaminations and/or adsorption problems. Lower detection limits could be achieved by increasing the deposition time or using a system to avoid interferences from the oxidant electrochemically produced at the auxiliary electrode, allowing use of lower pH and/or lower Edep.

Sb Determination in Natural Waters. Table 1 presents comparisons between Sb levels determined by voltammetry with literature values and values obtained by ICPMS for different water types: coastal, river, lake, mineral, and sparkling waters. Measurements were done in oxygenated water using 5 μm electrodes in DP mode (50 mV amplitude, 6 mV step, 100 ms interval time, 4 ms pulse time). Please note that no certified or literature value could be found for NASS-5; SLRS-5 and NASS-4 are compared to the information value (as no certified value is provided because of uncertainties in the error) and value found in the literature.31 SLRS-5 is analyzed in presence of 275 nM Cu. NASS-4 was kept in the laboratory for years and was contaminated with Hg. Sensitivities normalized to the length of the electrode and deposition time are indicated for each solution. Although straight comparison cannot be made due to different deposition potential and acidities, sensitivities tend to decrease in the order Milli-Q water, freshwater, and seawater which can partly be explained by the cation content (which affects the migration current) and the different chloride levels. Nevertheless, a sharp voltammetric peak is always obtained in these different matrixes (Figure 5), and detection of low nM levels are achieved with short deposition times (Table 1), proving the low detection limits achievable in these different matrixes by simply increasing tdep. Major discrepancy is however observed for the sparkling water which may be due to an overestimation of the Sb value obtained by ICPMS, as previously reported for As and Cr32 in presence of high levels of carbon dioxide. The determination of Sb in seawater is better achieved using background subtraction in acidic condition (0.1
0.3 M of HCl, HNO3, or H2SO4). Optimum experimental conditions in terms of measurement time, standard deviation, and long-term stability of the electrode were Edep =
1.8 to
2 V for 30
120 s, Ecleaning =
0.05 V (1
3 s), 1 s equilibrium time, stripping from
0.1 to 0.4 V (DP, 2
5 ms modulation time, 100 ms interval time, 50 mV amplitude, 4
6 mV step), Estandby = 0.55 V. If no cleaning potential was used, the Sb peak had a large shoulder on the negative potential side, which interfered with the peak measurement. Application of this cleaning potential caused the peak to be well shaped (e.g., Figure 5B). Three UV digested coastal water samples from Liverpool Bay collected in 3854

dx.doi.org/10.1021/ac200314q |Anal. Chem. 2011, 83, 3848–3856

Analytical Chemistry

ARTICLE

Figure 5. Typical examples of voltamograms obtained in (A) mineral water (analytical scan) and (B) seawater (background corrected). Addition of 2 nM Sb(V) in each case. Added electrolyte and deposition time given in Table 1.

June 2009 (using a metal free Niskin bottle) were analyzed with this method. Concentrations were 1.44 ( 0.09, 1.86 ( 0.16, and 1.18 ( 0.04 nM for salinities of 32.28, 32.84, and 33.17, respectively, which is in the expected range for coastal water samples.8

’ CONCLUSION The hydrogen wave has been setting the limit to voltammetric analysis due to interfering background current and hydrogen blockage of the electrode. This work has shown that it is possible to work at potentials far more negative than the onset of the hydrogen wave, with advantages related to increased sensitivity and selectivity. This expands the frontier of stripping voltammetry which should lead to new analytical methods for analytes that are poorly electroactive (e.g., Se or Al). This work was done with gold wire electrodes, but other metallic wire electrode materials should be tested. The effect and potential advantages of nascent hydrogen to either electrochemical hydride formation or catalytic effect are still very much unknown. For instance, preliminary experiments show that Pb is benefiting from such catalytic effect and could be reliably measured at very low levels on the gold electrode. Curious findings include the minimal interference of up to 30 mM sodium hydroxyde at potentials well below the reduction potential of sodium, suggesting possible protection by the nascent hydrogen. The possibility of the use of electrochemically produced hydrogen to improve mass transport is of an advantage for on-site analysis, as no stirring or vibrating devices are required. ’ ASSOCIATED CONTENT

bS

Supporting Information. Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT P.S. benefits from an EPSRC Advanced Fellowship (EP/ E061303) and acknowledges the EPSRC Life Science Interface (LSI) programme for financial support; K.G.-W. was supported by a

NERC research studentship. We thank Carmel Pinnington for the SEM pictures, Kalle Uroic and Joerg Feldman (Aberdeen
UK) for ICPMS analysis, and NOC Liverpool and the crew of RV Prince Madog for sample collection in Liverpool Bay.

’ REFERENCES (1) Prakash, R.; Srivastava, R. C.; Seth, P. K. Electroanalysis 2003, 15, 1410–1414. (2) Huang, H. L.; Jagner, D.; Renman, L. Anal. Chim. Acta 1988, 207, 37–46. (3) Huang, H. L.; Dasgupta, P. K. Anal. Chim. Acta 1999, 380, 27–37. (4) Salaun, P.; Planer-Friedrich, B.; van den Berg, C. M. G. Anal. Chim. Acta 2007, 585, 312
322, DOI: 10.1016/j.aca.2006.12.048. (5) Arimoto, R.; Duce, R. A.; Ray, B. J.; Tomza, U. Global Biogeochem. Cycles 2003, 17, DOI: 10.1029/2001gb001406. (6) Council Directive 98/83/EC, Official Journal of the European Communities, Publications Office of the European Union: Luxembourg, 3 November 1998. (7) Shotyk, W.; Krachler, M.; Chen, B. J. Environ. Monit. 2006, 8, 288
292, DOI: 10.1039/b517844b. (8) Filella, M.; Belzile, N.; Chen, Y. W. Earth-Sci. Rev. 2002, 57, 125–176. (9) Shotyk, W.; Krachler, M.; Chen, B.; Zheng, J. J. Environ. Monit. 2005, 7, 1238–1244. (10) (a) Renedo, O. D.; Martinez, M. J. A. Anal. Chim. Acta 2007, 589, 255
260, DOI: 10.1016/j.aca.2007.02.069. (b) Renedo, O. D.; Martinez, M. J. A. Electrochem. Commun. 2007, 9, 820
826, DOI: 10.1016/j.elecom.2006.11.016. (c) Adeloju, S. B.; Young, T. M.; Jagner, D.; Batley, G. E. Analyst 1998, 123, 1871–1874. (11) (a) Brihaye, C.; Gillain, G.; Duyckaerts, G. Anal. Chim. Acta 1983, 148, 51–57. (b) Gillain, G.; Brihaye, C. Oceanol. Acta 1985, 8, 231–235. (12) Postupolski, A.; Golimowski, J. Electroanalysis 1991, 3, 793–797. (13) (a) Tanaka, T.; Sato, T. J. Trace Microprobe Tech. 2001, 19, 521–531. (b) Tanaka, T.; Ishiyama, T.; Okamoto, K. Anal. Sci. 2000, 16, 19–23. (14) Lingane, J. J.; Nishida, F. J. Am. Chem. Soc. 1947, 69, 530–533. (15) (a) Gilbert, T. R.; Hume, D. N. Anal. Chim. Acta 1973, 65, 451–459. (b) Quentel, F.; Filella, M. Anal. Chim. Acta 2002, 452, 237–244. (c) Gillain, G.; Duyckaerts, G.; Disteche, A. Anal. Chim. Acta 1979, 106, 23–37. (16) Santos, J. R.; Lima, J.; Quinaz, M. B.; Rodriguez, J. A.; Barrado, E. Electroanalysis 2007, 19, 723–730. (17) Zong, P.; Nagaosa, Y. Microchim. Acta 2009, 166, 139
144, DOI: 10.1007/s00604-009-0176-9. (18) Chapman, C. S.; van den Berg, C. M. G. Electroanalysis 2007, 19, 1347
1355, DOI: 10.1002/elan.200703873. 3855

dx.doi.org/10.1021/ac200314q |Anal. Chem. 2011, 83, 3848–3856

Analytical Chemistry

ARTICLE

(19) Salaun, P.; van den Berg, C. M. G. Anal. Chem. 2006, 78, 5052–5060. (20) (a) Denkhaus, E.; Beck, F.; Bueschler, P.; Gerhard, R.; Golloch, A. Fresenius J. Anal. Chem. 2001, 370, 735–743. (b) Kuhn, A. T.; Byrne, M. Electrochim. Acta 1971, 16, 391–399. (21) Munteanu, G.; Dempsey, E.; McCormac, T. J. Electroanal. Chem. 2009, 632, 80
87, DOI: 10.1016/j.jelechem.2009.03.020. (22) Bolea, E.; Arroyo, D.; Cepria, G.; Laborda, F.; Castillo, J. R. Spectrochim. Acta, Part B: At. Spectrosc. 2006, 61, 96
103, DOI: 10.1016/j.sab.2005.12.001. (23) Salzberg, H. W.; Andreatch, A. J. J. Electrochem. Soc. 1954, 101, 528–532. (24) Gibbon-Walsh, K.; Salaun, P.; van den Berg, C. M. G. Anal. Chim. Acta 2010, 662, 1–8. (25) Meites, L. J. Am. Chem. Soc. 1954, 76, 5927–5931. (26) Bejan, D.; Bunce, N. J. J. Appl. Electrochem. 2003, 33, 483–489. (27) Huang, H. L.; Jagner, D.; Renman, L. Anal. Chim. Acta 1987, 202, 123–129. (28) Chen, Y. W.; Deng, T. L.; Filella, M.; Belzile, N. Environ. Sci. Technol. 2003, 37, 1163
1168, DOI: 10.1021/Es025931k. (29) Deng, T. L.; Chen, Y. W.; Belzile, N. Anal. Chim. Acta 2001, 432, 293–302. (30) Nash, M. J.; Maskall, J. E.; Hill, S. J. J. Environ. Monit. 2000, 2, 97–109. (31) Ding, W. W.; Sturgeon, R. E. J. Anal. At. Spectrom. 1996, 11, 225–230. (32) Pettine, M.; Casentini, B.; Mastroianni, D.; Capri, S. Anal. Chim. Acta 2007, 599, 191
198, DOI: 10.1016/j.aca.2007.08.016.

3856

dx.doi.org/10.1021/ac200314q |Anal. Chem. 2011, 83, 3848–3856