Calibrationless determination of cadmium, lead and copper in rain samples by stripping voltammetry at mercury microelectrodes

Analytica Chimica Acta 452 (2002) 65–75

Calibrationless determination of cadmium, lead and copper in rain samples by stripping voltammetry at mercury microelectrodes Effect of natural convection on the deposition step Mamdouh Elsayed Abdelsalam a , Guy Denuault a,∗ , Salvatore Daniele b a

b

Department of Chemistry, The University of Southampton, Highfield, Southampton SO17 1BJ, UK Department of Physical Chemistry, University of Venice, Calle Larga, S. Marta, 2137, 30123 Venice, Italy Received 15 May 2001; received in revised form 14 September 2001; accepted 27 September 2001

Abstract Mercury microelectrodes were prepared by ex situ deposition of Hg onto Pt microdiscs. By exploiting the known properties of microelectrodes in stripping analysis, an absolute method based on a simple equation derived from the stripping charge and the microelectrode steady-state current was assessed for the simultaneous quantification of Cd2+ , Pb2+ and Cu2+ concentrations. The method was tested with synthetic solutions containing known amounts of Cd2+ , Pb2+ and Cu2+ . Then, it was used to determine the labile and total fractions of these metal ions in rain samples. The labile fractions were measured from samples at their natural pH while the total concentrations were determined from samples at pH = 2. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Anodic stripping voltammetry; Mercury microelectrodes; Heavy metals; Rain water

1. Introduction Anodic stripping voltammetry (ASV) has always been regarded as one of the most sensitive techniques for trace metal analysis [1–6]. The very low detection limits achieved by ASV are due to the preconcentration of the analyte from the sample solution during the deposition step. Recent developments in sensor technology and micro-voltammetric electrodes [7–9] in particular, have extended the applicability of ASV to trace element analysis in complex and real samples. For instance, the fast ∗ Corresponding author. Tel.: +44-23-80592154; fax: +44-23-80593781. E-mail addresses: [email protected] (G. Denuault), [email protected] (S. Daniele).

steady-state mass transport by quasi-hemispherical diffusion obviates the need for stirring during the preconcentration step, improves the precision and reduces the analysis times. Small double layer capacitance resulting from the small electrode surface area significantly reduces the charging current. The ratio of faradaic to non-faradaic current is greater and the detection limit is enhanced [10]. Additionally, the small currents passed by microelectrodes result in negligible ohmic losses and this allows electroanalysis to be carried out in poorly conductive media [11,12] without the need for a supporting electrolyte. Impurities from the electrolyte are, thus, eliminated and microelectrodes are thus well-suited for ultratrace analysis. Furthermore, the metal is accumulated during the electrodeposition period in a very small volume of electrode, and

0003-2670/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 3 - 2 6 7 0 ( 0 1 ) 0 1 4 3 4 - 9

66

M.E. Abdelsalam et al. / Analytica Chimica Acta 452 (2002) 65–75

therefore, is completely reoxidised during the anodic scan. The unique properties of microelectrodes have led to the development of an attractive calibrationless method for quantification of trace elements by ASV. This method relies on a simple equation for the stripping charge and the steady-state current at the microelectrode. The later is assumed to remain constant during the deposition step [13,14]. For ASV experiments at mercury microelectrodes, the equation was derived under a further assumption that diffusion was the only mass transport operating during the deposition step. Measurements performed in low ionic strength natural samples may actually be affected by migration while natural convection phenomena may arise in prolonged experiments as in the ASV preconcentration step. In an earlier paper, it was shown that ASV responses recorded in low ionic strength aqueous solutions, were not affected by migration for divalent cations whose analytical concentration was below circa 0.1 ␮M [15]. On the contrary, to the best of our knowledge, no report exists on whether natural convection affects the deposition of metal ions during the deposition step for long times. This paper has therefore two objectives. The first is to show that natural convection does not affect the mass transport of the metal ion during the deposition step. The second is to assess the validity of the calibrationless equation for the detection of cadmium, lead and copper in low ionic strength rain samples with mercury microelectrodes. The validity of the calibrationless approach was tested on synthetic solutions containing concentrations of cadmium, lead and copper and using deposition times ranging from 300 to 1800 s, in order to assess the effect of natural convection. Subsequently, the method was applied to the determination of these heavy metals in rain samples.

2. Experimental 2.1. Instrumentation All voltammetric measurements were carried out with a two-electrode arrangement located in a solid aluminium Faraday cage. The reference electrode

was a home-made Hg/Hg2 SO4 electrode with saturated K2 SO4 solution. All potentials are referenced to this electrode. The potential of the microelectrode was controlled and programmed with a Hi-Tek Instruments PPR1 waveform generator. The current was amplified with a home-made current follower. The instruments were connected to a personal computer with an analogue to digital converter interface; the software controlling the system and the data acquisition was written in QuickBasic. The stripping charges were measured by integrating the area under the stripping peaks using the Microcal Origin program. 2.2. Mercury microelectrodes In this study, mercury microelectrodes were prepared by ex situ electrodeposition of mercury onto a freshly polished platinum microdisc of 10 ␮m diameter. The latter was fabricated by sealing a Pt microwire into glass and tested by voltammetry in ferrocyanide solution [16]. The deposition of mercury was performed from a solution (the plating solution) of 10 mM Hg2 (NO3 )2 and 1 M KNO3 at pH = 1 (acidified with HNO3 ), under potentiostatic conditions at high overpotential of −0.1 V against Hg/Hg2 SO4 saturated K2 SO4 reference electrode. Under the conditions chosen the mercury deposit should adopt a sphere-cap geometry. The height of the mercury deposit, h, was determined from the charge (Q) passed during the deposition step and the following equation [17] QM π h(3a 2 + h2 ) = 6 Fρ

(1)

where a is the radius of the microdisc, M and ρ the atomic mass and density of mercury, respectively. The mercury microelectrodes formed in this way were found to be durable and reproducible in their behaviour. The electrodes can normally be washed repeatedly with distilled water or left in solution while deoxygenating with a nitrogen stream without loss of the mercury. Steady-state voltammograms were recorded in 1 mM HClO4 and 0.1 M NaClO4 solutions to check whether the mercury was uniformly spread on the platinum microdisc. Well-behaved deposits caused a negative shift of the reduction wave for hydrogen

M.E. Abdelsalam et al. / Analytica Chimica Acta 452 (2002) 65–75

ions of about 1 V, owing to the high overpotential for hydrogen evolution on mercury compared to platinum [18]. At mercury microelectrodes the limiting current is given by [19] IL = KnFDcb a

(2)

where n is the number of electrons transferred, F the Faraday constant, D the diffusion coefficient of the electroactive species, cb the bulk concentration, a the radius of the microdisc electrode and the parameter K a geometric factor dependent on the ratio of the height of the sphere-cap to the radius of the substrate electrode, a. The coefficient K for the steady-state current equation at the various sphere-caps were also calculated by equation [19]
∞ cosh[x arctan(h/a)]dx K = 2π (3) 0 cosh[x arctan(a/ h)] cosh(π x/2) Electrode geometry and validity of Eq. (3) for the electrodes employed here were verified by comparing experimental and theoretical values of the parameter K. Experimental K values were derived from Eq. (2) and the current plateau recorded for H+ reduction at the mercury microelectrodes with different h/a ratios. All voltammograms were recorded in a solution of 1 mM HClO4 and 0.1 M NaClO4 at a sweep rate of 5 mV s−1 . A diffusion coefficient of 7.7×10−5 cm2 s−1 [20] was used for H+ . On the other hand, theoretical K values were obtained by substituting different ratios of h/a in Eq. (3). Both experimental and theoretical K values are reported in Table 1. General agreement within 3.5% was found. These

Table 1 Comparison of theoretical Kth values, calculated using Eq. (3), with experimental Kex values, measured by recording LSVs for H+ reduction with Hg microelectrodes of different h/a ratios h/a

Kex

Kth

100(Kex − Kth )/Kth

0.30 0.49 0.70 1.01 1.50 1.70 2.01

4.341 4.812 5.432 6.304 7.786 8.852 10.216

4.476 4.889 5.405 6.315 7.968 8.728 9.871

−3.02 −1.58 0.50 −0.17 −2.28 1.42 3.50

67

results agreed well with that reported in reference [13]. 2.3. Chemicals and samples Nitrate salts of Cd2+ , Pb2+ and Cu2+ (99.999% from Aldrich) were used to prepare stock standard solutions (1000 mg l−1 ) of these metal ions. Perchloric acid used to acidify the rain samples was of Aristar grade (BDH) and the other chemicals used were of Suprapur grade. All solutions were prepared using deionised water from a Millipore Milli-Q purification system. All samples were collected in polyethylene containers and purged with oxygen free nitrogen (BOC Ltd.) to get rid of dissolved oxygen. The nitrogen gas was presaturated with deionised water prior to purging to prevent evaporative losses from the sample solution. Prior to the first run, a purge time of 12–15 min was generally used to deoxygenate the sample solution. A home-made 10 ml Teflon cell was used in all stripping measurements. It was cleaned before the experiments using recommended procedures for trace analysis [21]. Blank anodic stripping experiments were also performed on Milli-Q water spiked with the amount of HClO4 employed for the acidification of the samples. No stripping peak discernible from the background was observed, even after a deposition time of 1800 s at −1.2 V. The rain samples were collected during December 1998 and January 1999 on the roof of the Chemistry Department, at the University of Southampton. All arrangements in cleaning the vessels and collecting the samples were met [22]. After measuring pH and conductivity, the samples were filtered through a clean cellulose acetate membrane filter with 0.45 ␮m pore size. Then the samples were divided into two parts; one of them was kept at its natural pH and the other was acidified with HClO4 to pH = 2 to prevent adsorption to container walls and to release complexed metal ions. The labile fraction of the metal ions was determined on the day of collection. For the total fraction, the samples were stored at 4 ◦ C up to the measurements. Before the stripping analysis, the samples were left to equilibrate at room temperature. The pH and the conductivity of the rainwater samples measured at room temperature immediately after collection are given in Table 2.

68

M.E. Abdelsalam et al. / Analytica Chimica Acta 452 (2002) 65–75

Table 2 pH, conductivity and Cl− concentration for rain samples Samples

pH

Conductivity (␮S cm−1 )

Cl− (mM)

I II III IV V

5.0 4.8 5.0 4.9 4.8

39.5 42.3 35.5 57.9 50.5

0.30 0.21 0.35 0.43 0.55

3. Results and discussion 3.1. Quantification of the trace elements concentration Fig. 1 shows a typical stripping voltammogram recorded at a mercury microelectrode (h/a = 1.5, a = 5 ␮m) in a quiescent solution of 2.5 × 10−7 M of both Cd2+ and Pb2+ and 2 × 10−7 M of Cu2+ in 0.1 M NaClO4 . The metal ions were deposited simultaneously at a potential of −1.2 V for 300 s. Linear sweep voltammetry (LSV) was applied in the stripping step. In this series of measurements an excess

of supporting electrolyte was employed to exclude migration from mass transport. As shown in Fig. 1, Ep values of −1.02, −0.82 and −0.41 V versus Hg/Hg2 SO4 , saturated K2 SO4 were obtained for the oxidation of Cd, Pb and Cu, respectively. The peak width at half-height, W1/2 , of about 38, 39 and 46 mV were obtained for Cd, Pb and Cu, respectively. These values are similar to those predicted for thin film behaviour [23,24]. For the quantification of Cd, Pb and Cu the following equation [13] was employed   (Ep − Ed ) b Qs = KnFDc a td + (4) ν where td is the deposition time, ν the scan rate, Ed the deposition potential and Ep the peak potential; other symbols have their usual meaning. The second term in the brackets represents the scanning period during which plating of the metal continues. Using Eq. (4) and the experimental charge Qs , the concentration cb of the analyte can be determined, provided that the other parameters are known. The reproducibility and precision of the anodic stripping measurements for the three metal ions were calculated from five

Fig. 1. ASV recorded at Hg microelectrode, h/a = 1.5, in a solution containing 2.5 × 10−7 M of both Cd2+ and Pb2+ and 2 × 10−7 M of Cu2+ in 0.1 M NaClO4 . E d = −1.2 V, t d = 300 s and sweep rate = 10 mV s−1 .

M.E. Abdelsalam et al. / Analytica Chimica Acta 452 (2002) 65–75

successive measurements in the same solution and under the same conditions that were used to record Fig. 1. The measurements were carried out using the same mercury microelectrode. The validity of the stripping charge approach was also checked by comparing the theoretically calculated stripping charges, (Qs )th , with that measured experimentally, (Qs )ex . The (Qs )th values were calculated using Eq. (4) after substituting the D values. The latter were calculated from D values at infinite dilution [25–27] and corrected for ionic strength [28]. The (Qs )ex values were obtained by integrating the stripping peaks of the voltammograms; all of these data are given in Table 3. It can be seen that relative standard deviations of 5.3, 2.4 and 3.1% were obtained for Cd, Pb and Cu, respectively. (Qs )th and (Qs )exp agree within 2.6%, for Cd and Cu and 1.6% for Pb, respectively, indicating the applicability of Eq. (4) to carry out the analysis. These results agree well with those reported earlier [13,14].

Table 3 Reproducibility of the stripping measurements and comparison of experimental and theoretical stripping charges using a mercury microelectrodea (Qs )ex (nC) Cd 0.43 0.39 0.38 0.40 0.42 Pb 0.6 0.62 0.61 0.62 0.64 Cu 0.38 0.41 0.38 0.39 0.39

Mean (Qs )ex (nC)

R.S.D. (%)

(Qs )th (nC)

Qs (%)

0.40

5.3

0.39

2.6

0.62

2.4

0.61

1.6

0.39

3.1

0.38

2.6

a (Q ) : the experimentally measured stripping charge; (Q ) : s ex s th the theoretically calculated stripping charge using Eq. (4); R.S.D.: the relative standard deviation; Qs = [((Qs )ex − (Qs )th )/(Qs )th ] × 100.

69

3.2. Effect of the preconcentration time and analyte concentration The effect of the preconcentration time, td , on the stripping charges is the main parameter to be investigated, as the length of the deposition step may give rise to natural convection phenomena. The effect of the deposition time was studied in a solution of 5 × 10−8 M of both Cd2+ and Pb2+ and 4 × 10−8 M of Cu2+ in 0.1 M NaClO4 . The three ions were preconcentrated simultaneously at a potential of −1.2 V and voltammograms were recorded at a potential scan rate of 10 mV s−1 . The preconcentration time was varied between 300 and 1800 s. The (Qs )ex versus preconcentration time plots for the three species are shown in Fig. 2. The regression analysis of these data yielded (Qs (nC)) = 0.002 + 3.42 × 10−3 (td (s)) with r 2 = 0.999 for Pb and (Qs (nC)) = 0.004 + 2.54 × 10−4 (td (s)) with r 2 = 0.997 for Cd. Moreover, (Qs )th , calculated using Eq. (4) and D values at a given ionic strength, and (Qs )ex agreed within 5% for Cd2+ and 2.8% for Pb2+ at the deposition times used. The linearity of the (Qs )exp versus td plots suggests that the effect of natural convection in the preconcentration step is either negligible or constant over the deposition times investigated. In fact, the onset of natural convection would have increased the rate of mass transport and a larger amount of metal would have accumulated in the mercury sphere-cap; accordingly, a larger stripping charge would have been observed. However, the relatively small discrepancies observed between experimental and theoretical data for cadmium and lead, at all deposition times, and for copper, at relatively low deposition times, allowed to exclude significant effects due to natural convection. For the Cu (see Fig. 2), the anomalous deviation from the linearity at longer preconcentration times can be interpreted on the basis of its low solubility in mercury (8 × 10−3 wt.% [29,30]). Exceeding this solubility limit may produce phenomena that has a detrimental effect on the subsequent stripping determination [31]. A separate crystalline phase of pure metal is often produced when the solubility is exceeded. To avoid these problems, the determination of Cu2+ was also performed separately after the determination of Cd2+ and Pb2+ . The deposition potential was chosen to selectively deposit the Cu2+ without depositing Cd2+ and Pb2+ and an appropriate deposition time

70

M.E. Abdelsalam et al. / Analytica Chimica Acta 452 (2002) 65–75

Fig. 2. Qs vs. td for the voltammograms recorded at a Hg microelectrode, h/a = 1.5 for a solution containing 5 × 10−8 M of both Cd2+ and Pb2+ and 4 × 10−8 M of Cu2+ in 0.1 M of NaClO4 . Sweep rate = 10 mV s−1 , E d = −1.2 V (䉱) Pb, (䊉) Cd and (䊏) Cu.

was used. When the above experiments were repeated following the suggested scheme; the same results were obtained for Cd and Pb, however, for Cu, deposited at Ed of −0.6 V, the linearity was extended up to a deposition time of 1800 s, (Qs (nC)) = 0.004+1.97×10−4 (td (s)) with r 2 = 0.996, also (Qs )th and (Qs )ex agreed within 6%. The dependence of Qs on the concentration of metal ions was also investigated for solutions containing different concentrations of Cd2+ , Pb2+ and Cu2+ in 0.1 M NaClO4 . Both Cd2+ and Pb2+ were deposited simultaneously and investigated first at a deposition potential of −1.2 V. Then, Cu2+ was deposited and investigated at Ed of −0.6 V. The voltammograms were recorded at a potential scan rate of 10 mV s−1 . A preconcentration duration of 300 s was used. The experimentally measured stripping charges were plotted against the concentration of the metal ions. Good linearity was obtained; the regression analysis of these data yielded (Qs (nC)) = 0.088+2.3 ([M2+ ]/(␮M)) with r 2 = 0.999 for Pb and (Qs (nC)) = 0.015 + 1.59 ([M2+ ]/(␮M)) with r 2 = 0.999 for Cd. Also the electrical charge associated with the stripping peaks agreed within 5.3 and 3.4% with

that predicted by Eq. (4) for Cd and Pb, respectively. For Cu a linear relationship between the plot of the stripping charge and the concentration was also obtained at all concentrations used. The regression analysis of these data yielded (Qs (nC)) = 0.017 + 1.4 ([M2+ ]/(␮M)) with r 2 = 0.997. Copper was selectively deposited in the mercury so as to reduce the concentration of the metals in the mercury electrode. Thus, the concentration of the Cu metal was under the solubility limits and the linearity was fulfilled. The experimentally measured stripping charge agreed within 6.2% with that predicted by Eq. (4). Thus, even in this case no effect due to natural convection was observed. 3.3. Analysis of rain samples Our aim is to test the applicability of the calibrationless approach with the mercury microelectrode for the direct measurement of trace metals in rain samples. The determination of the labile fraction was carried out first. From the conductivity values shown in Table 2 and on the basis of results reported earlier, no migration should arise in both the deposition and

M.E. Abdelsalam et al. / Analytica Chimica Acta 452 (2002) 65–75 Table 4 The optimum experimental conditions employed in the anodic stripping voltammetry analysis of rain samples Samples

I II III IV V

Cd

Pb

Cu

Ed (V)

td (s)

Ed (V)

td (s)

Ed (V)

td (s)

−1.2 −1.2 −1.2 −1.2 −1.2

1200 900 900 1800 1200

−0.95 −1.2 −0.95 −0.95 −0.95

300 900 600 900 600

−0.6 −0.6 −0.6 −0.6 −0.6

600 600 300 300 300

oxidation steps [15]. Moreover, from pH values, also included in Table 2, no or negligible adsorption of the metal ions onto the wall of the Teflon cell are expected to occur [32]. The experimental parameters Ed and td were optimised for each rain sample and metal ion. The Ed and td were adjusted according to the relative trace metal concentrations. These conditions were carefully investigated to achieve the best conditions with respect to signal to noise ratio, repeatability and speed of analysis. The optimised experimental parameters for all samples and metal ions are given in Table 4.

71

Typical stripping voltammograms recorded for rain sample II, are shown in Fig. 3. The W1/2 values and peak potentials obtained are also given in Table 6. If one compares these values with those recorded in the presence of supporting electrolyte, Fig. 1, it appears that the values are close to each other; this indicates, as expected, that the ohmic drop effects are minor. The W1/2 value changed from 39 to 43, 35 to 39, and 42 to 53 mV for Cd, Pb and Cu, respectively. These values are similar to those observed for the rain sample in reference [33] and to those obtained for a reversible two-electron process at conventional mercury film electrode [34]. Since the oxidation of copper can produce either the divalent or monovalent ion depending on the amount of chloride in the solution [35–37], the amount of chloride ions in the samples was evaluated by titration using standard procedures [38]. The chloride concentration was found to be less than 0.6 mM in all samples as reported in Table 2. In [39], the authors reported that the Cu stripping peak follows a two-electron oxidation up to 1 mM chloride ions. Beyond this chloride concentration, they observed a decrease in the stripping charge in addition to a dramatic shift of the peak

Fig. 3. Stripping voltammograms obtained at a Hg microelectrode, h/a = 1.6, for rain sample II. E d = −1.2 V for Cd2+ and Pb2+ and −0.6 V for Cu, t d = 900 s for Cd2+ and Pb2+ and 600 s for Cu2+ , sweep rate = 10 mV s−1 .

72

M.E. Abdelsalam et al. / Analytica Chimica Acta 452 (2002) 65–75

potential towards more negative values, and they attributed this shift to the stabilisation of Cu+ by Cl− and to the change from a two-electron to a one-electron oxidation. Thus, up to 1 mM chloride concentration the number of electrons involved in the copper oxidation can safely be assumed to be equal to two and they used this limit to analyse the Cu in rain samples. In

the present investigation, the chloride concentration is less than 0.6 mM and no significant shift in the peak potential for Cu was observed; thus, the number of electrons involved in the oxidation of Cu in the media investigated here was assumed to be equal to two. The use of Eq. (4) for the determination of the concentrations requires the knowledge of the diffu-

Fig. 4. Procedures for determining DCd in rain sample II, (a) LSVs recorded at Hg microelectrode, h/a = 2, after spiking the sample with (1) 5 ␮M, (2) 10 ␮M, (3) 15 ␮M and (4) 20 ␮M of the Cd2+ standard solution, sweep rate = 10 mV s−1 . (b) Id vs. [Cd].

M.E. Abdelsalam et al. / Analytica Chimica Acta 452 (2002) 65–75 Table 5 Diffusion coefficients determined in rain samples by recording linear sweep voltammograms at Hg microelectrodes Samples

DCd × 106 (cm2 s−1 )

DPb × 106 (cm2 s−1 )

DCu × 106 (cm2 s−1 )

I II III IV V

6.532 6.512 6.623 6.456 6.599

8.742 8.645 8.871 8.613 8.642

6.319 6.139 6.256 5.956 5.990

sion coefficient of the metal ions in the media investigated. In fact, the diffusion coefficients may change from one solution to another because of variations in ionic strength and viscosity. The D values were, therefore, determined in the following manner. An aliquot of the given sample was spiked with relatively large amounts of known concentrations of the investigated species and low scan rate linear sweep voltammograms were recorded with the mercury microelectrode. The steady-state limiting currents, IL , thus obtained were plotted against the metal ion concentrations. Fig. 4 shows typical steady-state voltammograms and the relevant plot for a series of measurements. According to Eq. (2), the plot should produce a straight line and the diffusion coefficient can be determined from the slope, as all the other terms are known. The D values thus obtained are listed in Table 5. One can recognise that the D values listed in Table 5 are not much different from their values at infinite dilution. This can be explained on the basis of the low ionic strength of rain samples as reflected by the low conductivity values recorded in Table 2. Finally, Qs and D values were substituted in Eq. (4) to calculate the labile fraction of the trace elements in the rain samples; the results obtained are reported in Table 6. Each value reported in this table is the mean of five replicate measurements; in each case the relative standard deviation is included. The total concentrations of the metal ions were determined from the acidified samples. Both the stripping charge and the standard addition methods were used to determine the ion concentrations. For the stripping charge method, stripping voltammograms were recorded in rain samples under conditions similar to that reported in Table 4. Eq. (4) was then used to calculate the total concentration of the species. The diffusion coefficients were calculated

73

Table 6 Results obtained for the analysis of rain samples at natural pH using the stripping charge approach Samples I

II

III

IV

V

Cd [Cd2+ ] (␮g l−1 ) R.S.D. (%) Ep (V) W1/2 (mV)

0.5 5.6 –1.01 41

0.7 5.1 –1.03 39

0.9 6.7 –1.0 42

0.3 7.1 –1.01 41

0.8 6.9 –1.02 43

Pb [Pb2+ ] (␮g l−1 ) R.S.D. (%) Ep (V) W1/2 (mV)

7.1 2.5 −0.80 35

2.0 3.1 −0.82 36

5.9 4.4 −0.79 37

2.6 3.5 −0.81 35

3.8 5.2 −0.80 39

Cu [Cu2+ ] (␮g l−1 ) R.S.D. (%) Ep (V) W1/2 (mV)

2.7 4.5 −0.40 52

3.3 4.2 −0.41 42

3.7 4.8 −0.39 50

3.3 5.2 −0.39 49

3.3 5.9 −0.4 53

from values at infinite dilution and corrected for the conductivity of the rain samples and 0.01 M ionic strength of HClO4 . The results of the analysis are reported in Table 7. This table also includes the values obtained by the standard addition method. Table 7 Comparison between the stripping charge (SC) and the standard addition (SA) methods for determination of total Cd, Pb and Cu concentrations in acidified rain samples by ASV Samples I

II

III

IV

V

Cd (␮g l−1 ) SC R.S.D. (%) SA R.S.D. (%)

0.8 5.3 0.9 5.1

1.1 4.2 1.0 3.5

1.1 6.5 1.0 3.6

0.7 6.0 0.7 5.54

0.9 6.6 0.8 6.4

Pb (␮g l−1 ) SC R.S.D. (%) SA R.S.D. (%)

9.5 3.5 9.2 2.7

3.1 5.9 2.8 5.0

7.4 4.1 7.3 2.9

6.4 4.7 6.5 3.8

7.2 4.6 7.0 2.5

Cu (␮g l−1 ) SC R.S.D. (%) SA R.S.D. (%)

5.0 4.4 5.1 3.5

4.8 3.6 4.8 2.9

6.2 5.0 5.7 5.1

5.7 3.5 5.7 5.2

4.4 6.2 3.9 5.6

74

M.E. Abdelsalam et al. / Analytica Chimica Acta 452 (2002) 65–75

The precision of the determination was tested and the relative standard deviation was calculated. The latter was established on the basis of three replicate measurements. The means of the two sets of measurements were tested to see if they were significantly different or whether the discrepancy was simply a consequence of indeterminate errors in the two sets of measurements. This was achieved by carrying out the significance test (t-test) [40]. The difference between the two means is found to be less than the critical value of t, i.e. 2.78, at the significance level of 95% (i.e. probability, P = 0.05). This result indicates that the null hypothesis is retained and the observed difference between the concentrations determined by both stripping charge and standard addition methods is due to indeterminate errors, i.e. no determinate errors are demonstrated. Considering the data shown in Tables 6 and 7, it is evident that the total concentrations of the metal ions determined in the acidified samples are greater than that found in the untreated samples. This difference could result from the breakdown of the metal-complexing organic ligands that occur naturally in the rain samples. The difference between total and labile concentrations is strongly related to the speciation character of the metal ion in aquatic systems [41,42]. In order to exclude any interference from the release of metal ions potentially adsorbed onto the wall of the sample containers, blank experiments were also carried out on Milli-Q water samples spiked with NaCl, HClO4 and metal ions to achieve the acidity, chloride ions and heavy metals levels similar to those obtained experimentally in the rain samples. The measurements were performed on freshly prepared solutions and 24 h after their preparation. The concentration values thus determined did not differ by more than 5%; that is within or close to the R.S.D. (%) reported in Tables 6 and 7. Since there are no published data for the Southampton area, the data found in this paper were compared with those found in places with similar geographical characteristics, that is large seaside towns with significant industrial activities. Thus, the concentration level of Cl− found here is close to the maximum values found in Higashi, Hiroshima (Japan) [43] and Venice (Italy) [39]. The rather high Cl− concentrations found are probably related to the proximity of the sea. Similarly, the heavy metal concentrations are close to the maximum values found in those areas and are thought

to be linked to the proximity of large industrial complexes. Concentration levels in rain strongly depend on the season, on whether samples are collected at the start or end of the rainfall and on the number of atmospheric events [43]. These aspects were not considered here, as it was beyond the scope of this study.

4. Conclusion The results reported here have shown that sphere-cap mercury microelectrodes can be employed for the determination of metal ions by ASV, using a calibrationless procedure. Although relatively long deposition times have been employed during the deposition step, no effect due to natural convection has been observed. Moreover, accurate and reproducible data could be obtained in rain samples at low ionic strength under natural conditions. The calibrationless method employed here to determine both the labile and total fractions of lead, cadmium and copper, should speed up the routine determination of these metal ions by ASV. References [1] Kh. Brainina, E. Neyman, Electroanalytical Stripping Methods, Wiley, New York, 1993. [2] J. Wang, Stripping Analysis: Principles, Instrumentation and Applications, VCH Publishers, Deerfield Beach, 1985. [3] F. Vydra, K. Stulik, E. Juláková, Electrochemical Stripping Analysis, Ellis Horwood, Chichester, 1976. [4] Kh. Brainina, Stripping Voltammetry in Chemical Analysis, Halsted, New York, 1974. [5] R. Kalvoda, Electroanalytical Methods in Chemical and Environmental Analysis, Plenum Press, New York, 1987. [6] P.T. Kissinger, W.R. Heineman (Eds.), Laboratory Techniques in Electroanalytical Chemistry, Marcel Dekker, New York, 1984. [7] M. Fleischmann, S. Pons, D.R. Rolison, P.O. Schmidt (Eds.), Ultramicroelectrodes, Datatech Systems, Morganton, NC, 1987. [8] R.M. Wightman, D.O. Wipf, in: A.J. Bard (Ed.), Electroanalytical Chemistry, Marcel Dekker, New York 15, 1988, p. 267. [9] M.I. Montengro, M.A. Queiros, J.L. Daschbach, Microelectrodes: Theory and Applications, NATO ASI Series, Vol. E197, 1991. [10] K.R. Wehmeyer, M.R. Deakin, R.M. Wightman, Anal. Chem. 57 (1985) 1913. [11] S. Daniele, M.A. Baldo, P. Ugo, G.A. Mazzocchin, Anal. Chim. Acta 219 (1989) 9.

M.E. Abdelsalam et al. / Analytica Chimica Acta 452 (2002) 65–75 [12] D.K.Y. Wong, A.G. Ewing, Anal. Chem. 62 (1990) 2697. [13] M.A. Baldo, S. Daneile, M. Corbetta, G.A. Mazzocchin, Electroanalysis 7 (1995) 980. [14] M.A. Baldo, C. Bragato, S. Daniele, Analyst 122 (1997) 1. [15] S. Daniele, G.A. Mazzocchin, Anal. Chim. Acta 273 (1993) 3. [16] R.M. Penner, M.J. Heben, N.S. Lewis, Anal. Chem. 61 (1989) 1630. [17] C.L. Colyer, D. Luscombe, K.B. Oldham, J. Electroanal. Chem. 283 (1990) 379. [18] R.N. Adams, Electrochemistry at Solid Electrodes, Marcel Dekker, New York, 1969. [19] J.C. Myland, K.B. Oldham, J. Electroanal. Chem. 288 (1990) 1. [20] R. Mills, V.M.M. Lobo, Self Diffusion in Electrolyte Solutions, A Critical Examination of Data Compiled from the Literature, Physical Sciences Data, Vol. 36, Elsevier, Amsterdam, 1989. [21] L. Mart, Fresenius’Z. Anal. Chem. 296 (1979) 350. [22] M. Kopanica, F. Opekar, in: R. Kalvoda (Ed.), Electroanalytical Methods in Chemical and Environmental Analysis, Plenum Press, New York, 1987. [23] K.R. Wehemeyer, R.M. Wightman, Anal. Chem. 57 (1985) 1989. [24] W.T. De Vries, E. Van Dalen, J. Electroanal. Chem. 14 (1967) 315. [25] D.E. Gray (Ed.), American Institute of Physics Handbook, McGraw-Hill, New York, 1972, pp. 2–226. [26] R.A. Robinson, R.H. Stokes, Electrolyte Solutions, Butterworths, London, 1959.

75

[27] V.M.M. Lobo, J.L. Quaresma, Handbook of Electrolyte Solutions, Physical Science Data, Series 41, Elsevier, Amsterdam, 1989. [28] J.O. Bockris, A.K.N. Reddy, Modern Electrochemistry, Vol. 1, Plenum Press, New York, 1970. [29] A.F. Trotman-ickenson (Ed.), Comprehensive Inorganic Chemistry, Vol. 3, Pergamon, London, 1973 p. 283. [30] Z. Galus, Crit. Rev. Anal. Chem. 4 (1975) 359. [31] G. Jangg, H. Kirchmayr, Z. Chem. 3 (1963) 47. [32] S. Diaz-Cruz, J.M. Diaz-Cruz, M. Esteban, Anal. Chem. 66 (1994) 1548. [33] S. Daniele, M.A. Baldo, P. Ugo, G.A. Mazzocchin, Anal. Chim. Acta 219 (1989) 19. [34] C. Brihaye, G. Duyckaerts, Anal. Chim. Acta 146 (1983) 37. [35] M.L.S. Simoes Goncalves, M.M. Correira Dos Santos, J. Electroanal. Chem. 143 (1983) 397. [36] M.M. Correira dos Santos, M.L. Simoes Goncalves, Electroanalysis 3 (1991) 131. [37] S. Daniele, M.J. Pena, Electrochim. Acta 38 (1993) 165. [38] I.M. Kolthoff, P.J. Elving (Eds.), Treatise on Analytical Chemistry, Part 1, Vol. 3, Wiley, New York, 1979, p. 349. [39] S. Daniele, C. Bragato, M.A. Baldo, Anal. Chim. Acta 346 (1997) 145. [40] J.C. Miller, J.N. Miller, Statistics for Analytical Chemistry, 2nd Edition, Ellis Horwood, UK, 1988. [41] J.W. Moore, S. Ramamoorthy, Heavy Metals in Natural Waters, Applied Monitoring and Impact Assessment, Springer, Berlin, 1984. [42] T.M. Forence, Analyst 3 (1986) 489. [43] K. Takeda, K. Marumoto, T. Minamikawa, H. Sakugawa, K. Fujiwara, Atm. Environ. 34 (2000) 4525.