Introduction

The determination of trace amounts of cobalt in natural waters is of great interest because the cobalt, as a major constituent of vitamin, is important for humans and other living organisms. In addition, cobalt-60, a radioactive isotope of cobalt, is an important source of gamma rays which is often used for treating some cancer or as a medical tracer. However, toxic effects of the excess cobalt have been found, which include polycythemia, skin allergies, pulmonary disorders, and other adverse reactions [1]. In 1998, the European Union identified the cobalt chloride as a carcinogen in the second category. The production and usage of blue silica gel which contains cobalt chloride has been banned in Europe and the USA. The water contaminated by cobalt became a big threat to not only the ecological environment system but also the human health. Hence, monitoring and determination of cobalt is highly urgent in many fields, especially in water system.

The determination of cobalt in water samples often requires sensitive methods and portable instrument, because cobalt is often at a low concentration [2, 3]. However, the common methods including chemiluminescence detection [4], electrothermal atomic absorption spectrometry [5], flame atomic absorption spectrometry [6], graphite furnace atomic absorption spectrometry (GFAAS) [7], and microwave plasma torch–atomic emission spectrometry [8] often do not have enough sensitivity, selectivity, and freedom from matrix interferences. These methods must combine with pretreated procedures. In contrast, the electrochemical stripping analysis has been recognized as a typical approach for trace metal ion measurement because it combines the advantages of the high speed, high sensitivity, simple equipment requirements, as well as low expenditure for tests and maintenance. Besides, it also supplies a possibility of constructing useful and portable apparatuses for on-site environmental monitoring.

Unlike the anodic stripping voltammetry (ASV), the adsorptive cathodic stripping voltammetry (AdCSV) has often been used for the determination of trace amounts of Co, Ni, and Cr which can neither form amalgams with mercury nor form metal membrane in the inert metal electrodes. The AdCSV determination usually involves the formation of metal complexes and an accumulation on the working electrodes and then the subsequent measurement of the reduction or oxidation peak of the accumulated complex of the electrode surface. Various complex agents have been used to determine the cobalt including dimethylglyoxime (DMG) [9–11], nioxime [12–15], α-furil dioxime [2], and pyrogallol red [16]. Among them, DMG and nioxime are the most popular complex agents because they offer the higher selectivity, the better shape of voltammetric signal, and the more excellent sensitivity.

The proper choice of the working electrode is very important for the electrochemical analysis. The mercury (Hg) electrodes not only have a highly reproducible and easily renewable surface but also have a high hydrogen overvoltage [17]. However, its use has been banned in many countries due to its hyper-toxicity. In last decades, various electrodes have been investigated for substituting Hg electrode, such as glassy carbon electrode, graphite electrode, gold electrode, silver electrode, and iridium electrode. But none of these electrodes has been found to be acceptable for their relatively low hydrogen overvoltage (usually below −0.8 V [18]), large background current, or poor precision and resolution [19].

In 2000, bismuth-based electrodes were introduced as a favorable alternative mercury-free electrode for stripping voltammetry analysis of trace metals [20]. It offers the closest behavior to mercury but with very low toxicity. Both in situ or ex situ bismuth film electrode [21, 22], bismuth bulk electrode (BiBE) [23, 24], and tin–bismuth film electrode [25] display an attractive stripping behavior. However, the bismuth bulk electrode is complicated to be fabricated due to the fragileness of the bismuth. For the bismuth film and tin-bismuth film electrode, when they are used in AdSV analysis where alkaline solutions are typical supporting electrolytes, it is inconvenient because of the risk of hydrolysis of Bi3+ species [23, 26] which limits its practical applications. Then, some concerns were focused on lead film electrode [27–29] and lead–copper film electrode [30]. Lead can be stabilized easily at mild alkaline pH, but it is widely known as a toxic metal. In addition, the preparation of those modified film electrodes is complex, which make the electrodes lack of consistency and mechanical strength. Considering both the advantages and disadvantages of aforementioned electrodes, the tin–bismuth alloy electrode (SnBiE), which has been proposed in our work [31–33], is used in AdCSV determination of cobalt in this paper. This new SnBiE based on the tin–bismuth alloy is of little environmental damage, and it is conveniently fabricated and used. Besides, it possesses a high hydrogen overvoltage which expands the scope and applicability of electrochemical technique for the determination of metal ions at negative potentials [33].

Here, in this work, for the first time, SnBiE was successfully used for the detection of trace amount of cobalt ions in the presence of nitrite in combination with an adsorptive cathodic stripping voltammetry. Nioxime acts as a complexing agent. It was found that the optimal sensitivity can be achieved by using nitrite as the auxiliary reagent, which allows a detection limit of sub-nanomolar level without any deoxygenating or electro-plating pretreatment. A comparison of the applicable potential windows was conducted at the same pH between the BiBE and the SnBiE. In addition, the compositions of the supporting electrolyte and the analytical parameters (i.e., deposition time and deposition potential) have been optimized. Based on the established method, the practical applications have been carried out for the determination of Co2+ in tap water samples.

Experimental

Reagents

Sn–Bi alloy wires (Sn:Bi, 42:58 wt%, 1 mm in diameter) were purchased from KAIT Electronic Material Co., Ltd., Guangzhou, China. A stock solution of cobalt(II) was prepared by dilution of the National Standard Substances of Cobalt (GBW(E)081005, 1 mg/mL), acidified to pH of 2.0 with distilled HCl, and then diluted to various concentrations of working solutions. An aqueous stock solution of 0.05 M nioxime (1,2-cyclohexanedione dioxime) was prepared by dissolving the appropriate amount in 0.2 M sodium hydroxide. A 0.1 M dimethylglyoxime was prepared by diluted into absolute ethyl alcohol, and 0.5 M sodium nitrite was prepared by diluted into ultrapure water. All of those solutions were prepared weekly. 0.1 M ammonia buffer solution (pH 9.4) prepared with ammonium chloride and ammonia was used as the supporting buffer solution for Co2+ determination. All other chemicals were analytical reagents and used without further purification. Ultrapure water (18.2 MΩ cm specific resistance) obtained from a Pall Cascada laboratory water system was used throughout the experiments. All of electrochemical experiments were carried out at room temperature and under a non-deoxygenating condition.

Apparatus

All electrochemical experiments were carried out in a conventional three-electrode cell controlled by a CHI 660D Electrochemical Work Station (CH Instruments, Inc.). A Sn–Bi alloy disk electrode (SnBiE) was used as the working electrode. A platinum foil was applied as the counter electrode, and a saturated calomel electrode (SCE) served as the reference electrode. The SnBiE was fabricated with a Sn–Bi alloy wire (2 cm in length, 1 mm in diameter) which was conducted with a copper wire and encapsulated in Teflon to make a disc electrode. The pH measurements were performed at an E-201-C model pH meter (Shanghai Leici Instrument Factory). Inductively coupled plasma–mass spectrometry (ICP-MS) was measured with an ELAN DRC II inductive coupling plasma–mass spectrometer (Perkin Elmer Ltd., Hong Kong).

Analytical procedure

The analysis of Co2+ was performed in a 25-mL beaker containing 20 mL 0.1 M ammonia buffer solution without removal of oxygen. Prior to analysis, the SnBiE was polished with a #2500 silicon carbide abrasive paper for a smooth and bright surface, and it did not need any activation. For Co2+ determination, the accumulation of Co2+ was carried out at −1.0 V (vs SCE) under stirring with an accumulation time of 60 s. The stripping voltammogram was then recorded in quiescent solution from −1.0 to −1.2 V (vs SCE) after a 10-s quiescence period.

Parameters of DP-AdCSV were as follows: step increment, 4 mV; pulse amplitude, 50 mV; pulse width, 0.025 s; sample width, 0.0167 s; pulse period, 0.05 s.

Results and discussion

The potential window of the SnBiE

A comparison of the applicable potential windows was conducted between the SnBiE and bismuth bulk electrode (BiBE) in 0.1 M ammonia buffer (pH 9.4) (Fig. 1). As shown in Fig. 1, the BiBE exhibits a relatively limited hydrogen overvoltage (in the vicinity of −1.40 V) but a wide anodic range (up to −0.40 V) with an applicable potential window of 1.00 V, which is in good agreement with the behavior observed by Joseph Wang [34] and A. Economou [35]. Compared to the BiBE, a wider cathodic potential range of SnBiE is found up to −1.60 V, though a limit of anodic potential region is found to be around −0.85 V due to the oxidation of Sn. Therefore, SnBiE with an applicable negative potential window of up to −1.60 V possesses a higher hydrogen overpotential than the BiBE, which indicates that it is extremely suitable for the adsorptive cathodic stripping voltammetry (AdCSV).

Fig. 1
figure 1

The accessible potential windows of BiBE (black line) and SnBiE (red line) in 0.1 mol/L ammonia buffer solution (pH 9.4). Linear sweep voltammetry; scan rate, 100 mV/s

Full size image

Enhancement effect of nitrite on the Co(II)–nioxime

In this work, the catalytic effect of nitrite ions was studied on the Co(II)–nioxime cyclic voltammetric response at the SnBiE. Cyclic voltammograms after the addition of nioxime and nitrite in 0.1 M ammonia buffer (pH 9.4) containing 0.2 μM Co(II) were recorded (Fig. 2). As can be seen from Fig. 2, in the potential range from −1.3 to −1.0 V, the cyclic voltammograms show that there is a low background current and no redox peak with the addition of 0.2 μM Co2+ solution (curve a). After the introduction of nioxime into the solution, a Co(II) cathodic reduction peak becomes apparent as a result of adsorptive accumulation and further reduction of the Co(II)–nioxime complex on the SnBiE surface (curve b, i p  = 1.251 × 10−6 A). However, no oxidation peak was observed at SnBiE on the reverse scan of the cyclic voltammograms, which elucidated that the electrode reaction was an irreversible charge transfer process. The subsequent addition of nitrite ions caused an enhancement of the Co(II)–nioxime cyclic voltammetric response (curve c, i p  = 3.013 × 10−6 A), which is the result of catalytic reduction of nitrite ions. The SnBiE exhibits a well-defined electrochemical behavior in the negative potential region with low background and high electrochemical response, which demonstrates that the alloy electrode possesses good interfacial electrochemical characteristics and implies the possibility of voltammetric detection of Co2+ on this SnBiE.

Fig. 2
figure 2

Cyclic voltammograms at SnBiE in (a) 0.1 M ammonia buffer solution (pH 9.4) containing 0.2 μM Co2+ ions, (b) as (a) + 200 μM nioxime, and (c) as (a) + 200 μM nioxime + 0.5 M nitrite ions. Scan rate, 50 mV/s

Full size image

The mechanism for adsorption cathodic stripping voltammetry of Co(II)–nioxime complex on the SnBiE can be explained according to the Hg electrodes which was proposed by Andrzej Bobrowski et al. [36] and Marisol Vega et al. [12]:

Firstly, in the presence of nioxime, ammonia ions, and nitrite ions, the Co2+ is oxidized to Co3+ in the solution, and then it formed a complex of CoIII(nioxime)2NO2NH3 near the electrode surface.

$$ {Co}^{2+}+2\ \mathrm{nioxime}+{NH}_3+{{\mathrm{NO}}_2}^{-}\to {Co}^{3+}{\left(\mathrm{nioxime}\right)}_2{\mathrm{NO}}_2{NH}_{3\mathrm{sol}}+{\mathrm{e}}^{-} $$
(1)

Then, during the accumulation period at an accumulation potential of −1.0 V, the CoIII(nioxime)2NO2NH3 complex is absorbed onto the SnBiE surface.

$$ {Co}^{3+}{\left(\mathrm{nioxime}\right)}_2{\mathrm{NO}}_2{NH}_{3\mathrm{sol}\ }\to {Co}^{3+}{\left(\mathrm{nioxime}\right)}_2{\mathrm{NO}}_2{NH}_{3\mathrm{ads}}\left({\mathrm{E}}_{acc}=-1.0\ \mathrm{V}\right) $$
(2)

Finally, followed by a negative voltammetric scan, the CoIII(nioxime)2NO2NH3 complex was reduced to the Co(II) complex with a one-electron transfer process at −1.128 V. The reduced species desorb from the electrode surface, followed by diffusion into the double layer where it is reoxidized to Co(III) by nitrite. The reduction current is catalytically enhanced by the presence of nitrite, which is represented by the Eq. (3):

(3)

Optimization for determination of Co2+ at the SnBiE

Comparison of different ligands

Nioxime and DMG ligands are usually used in the determination of cobalt at different electrodes. Therefore, we mainly compared the two ligands to evaluate their sensitivity for cobalt. The ligand concentrations were changed to determine the optimal ligand concentration for maximum sensitivity in 0.1 M ammonia buffer (pH 9.4) spiked with 2 nM cobalt(II), and the adsorptive cathodic stripping voltammograms were recorded from −0.90 to −1.20 V after a 60-s adsorption at −0.90 V. The optimized concentration of the ligands was determined by the highest increase of peak current when equivalent ligand was added to the solution. A linear analysis was applied to detect the sensitivity of the ligands for the determination of cobalt. The optimum ligand concentration, the sensitivity, and the peak potential for each ligand test are listed in Table 1. As shown in Table 1, the sensitivity of nioxime was 126 times more than that of DMG, but the reagent consumption was only one fiftieth of that of DMG. The peak potential for cobalt in the presence of nioxime was slightly more positive than DMG. The largest sensitivity was obtained using nioxime, which had the highest sensitivity but with the less reagent consumption. Therefore, nioxime was selected as complex ligand.

Table 1 Comparison of optimized conditions and sensitivity for the determination of cobalt using various ligands
Full size table

Effects of the solution composition

Three buffer solutions including ammonia buffer, phosphate buffer, and borate buffer solution were chosen for study of their suitability for Co2+ determination at the SnBiE. The voltammograms (not shown) indicated that the determination sensitivity was poor, and the catalytic effect was absent unless the ammonia buffer was present. Stable and sensitive signals were obtained for the ammonia buffer solution and it was, therefore, used in subsequent study. Furthermore, different concentrations of ammonia buffer solution, in the range of 0.01–0.10 M, were tested at a constant concentration of Co2+ (2 nM). The best performance can be obtained in 0.10 M ammonia buffer solution due to its better current response. Therefore, 0.10 M ammonia buffer solution was selected for further experiments.

Effects of the nioxime concentration

To determine the optimal ligand concentration for the maximum sensitivity, the effect of nioxime concentration was examined from 0 to 100 μM. Voltammograms were recorded from −0.90 to −1.20 V after a 60-s accumulation at −0.90 V (Fig. 3a). Results showed that the peak current for cobalt increased rapidly with the increase of nioxime concentration up to 60 μM and then increased slowly with the increase of nioxime concentration up to 100 μM due to the achievement of the maximum coverage. The increase of the nioxime concentration also caused the appearance of a broad reduction peak at −1.0 V, which is presumably due to the reduction of the ligand. Therefore, an optimum nioxime concentration of 60 μM was selected for further experiments.

Fig. 3
figure 3

Effect of varying a nioxime concentration and b nitrite concentration on the AdCSV sensitivity for 2 nM Co2+ at the SnBiE. Accumulation potential, −1.4 V; accumulation time, 60 s; equilibration time, 4 s; increment potential of each step, 5 mV; pulse amplitude, 25 mV

Full size image

Effect of the nitrite concentration

The peak current of cobalt-nioxime complex as a function of the auxiliary reagent (nitrite) concentration is displayed in Fig. 3b. The concentration of nitrite varies from 0 to 0.45 M. It can be found that the peak height for cobalt increases linearly with the nitrite concentration until the concentration is up to 0.3 M. Then, it levels off at higher concentrations. Meanwhile, the peak of cobalt was found to shift toward more negative potential as the nitrite concentration increased, which suggested that the formation of cobalt complexes was more stable. Therefore, 0.3 M nitrite concentration was selected for further experiments.

Effect of pH values

The effect of pH of the supporting electrolyte was studied by varying the pH in the range of 8.0 to 10.0. The peak current of cobalt as a function of pH is shown in Fig. 4a. The best sensitivity of the electrode for the Co2+ measurement in the absence of interfering ions was obtained at pH 9.4 because it produced maximum sensitivity for cobalt with small interference by the free nioxime. At pH lower than 9.4, the reduction peak of nioxime increased gradually as pH decreased which interfered with the stripping peak of Co2+. As for pH higher than 9.4, the response current decreased significantly, probably as a result of hydrolysis of cobalt ions at increasing pH. Therefore, ammonia buffer solution with pH 9.4 was selected as the optimum experimental condition.

Fig. 4
figure 4

Effect of varying a pH values of buffer solution, b accumulation potential, and c accumulation time on the stripping peak current of 2 nM Co2+ at the SnBiE. Other conditions are the same as in Fig. 3

Full size image

Effect of accumulation potential

The effect of accumulation potential on the stripping peak current of 2 nM Co2+ was examined over the potential range of −0.85 to −1.05 V. As shown in Fig. 4b, the peak current increases greatly with changing potential from −0.85 to −1.00 V, probably due to the fact that Co2+ can be reduced more efficiently at more negative deposition potentials. Experiments also showed that the peak current decreased at a potential more negative than −1.00 V, and a higher background current was apparent which might be attributed to the co-hydrogen evolution at such negative potentials. Therefore, an optimum deposition potential of −1.00 V was used in the following experiments.

Effect of accumulation time

The effect of the accumulation time on the peak current of Co2+ was investigated. The corresponding results are shown in Fig. 4c. The variation of the peak current is depicted with the accumulation time for 2 nM Co2+ in the range from 30 to 150 s. The response current increases rapidly with the accumulation time up to 150 s and then increases slowly. Though a longer accumulation time can increase the amount of Co2+ on the electrode surface and improve the sensitivity of practical determination, a short analysis time of 60 s is sufficient for determination of Co2+ in tap water. Additionally, for determination of lower concentration of Co2+ in other matrixes, a prolonged accumulation time is recommended.

Calibration curve

The calibration curve of Co2+ determination at the SnBiE by using AdCSV is shown in Fig. 5. For a 60-s accumulation, a proportional relationship between the reduction peak current and the concentration of Co2+ was obtained in the range of 0.2 to 20 nM by fitting the following regression equation:

$$ {i}_p=11.91+43.24C,r=0.998 $$
Fig. 5
figure 5

Calibration curve of Co2+ at the SnBiE. The concentrations of Co2+ for the inset curves are 0.2, 1, 3, 5, 7, 9, 15, and 20 nM from the bottom to the top, which are in the linear range. Other conditions are the same as in Fig. 3. (When not visible, error bars are smaller than the symbols)

Full size image

where i p is the reduction peak current in nA, and C is the concentration of Co2+ in nM. The sensitivity of the SnBiE to Co2+ is 43.24 nA/nM. The detection limit was given by the equation C L  = 3S bl/S, where S bl is the standard deviation of the blank measurements and S is the sensitivity of the calibration graph. The detection limit of Co2+ for 60 s accumulation was calculated to be 44 pM, which is lower than that at hanging mercury drop electrode [2, 16], stationary mercury drop electrode [37], cylindrical silver-based renewable amalgam film electrode [13, 38], in situ and ex situ plated bismuth film electrode [39–42], solid bismuth disk vibrating electrode [43], in situ plated lead film glassy carbon electrode and screen-printed electrode [9, 27, 28], and in situ plated lead–copper film electrode [30]. The data for the application of various electrodes in the AdCSV determination of Co2+ is show in Table 2.

Table 2 The data for the application of various electrodes in the adsorptive stripping voltammetry of Co(II)
Full size table

Six SnBiEs for 2 nM of Co2+ were estimated and the results revealed that the SnBiE had satisfactory reproducibility with a relative standard deviation of 3.5 %. Furthermore, to ensure the repeatability of the electrode, a series of seven repetitive measurements of 5 nM Co2+ with a relative standard deviation of 2.6 % were obtained. As a new electrode with a high repeatability, SnBiE is a potential for the practical analysis.

Study of interferences

The effect of possible interfering species was studied by analyzing a standard solution (2 nM Co2+) to which amounts of interfering ions were added under the optimized conditions (Fig. 6). Interfering ions were added at different concentrations (max. 1000-fold) which are higher than the concentration of Co2+ until they produced a change of ≥5 % of the initial height. The results show that more than a 1000-fold excess of Ni2+, Cu2+, Mn2+, Zn2+, Pb2+, Cr3+, Fe3+, Ca2+, and Mg2+ and less than 50-fold Cd2+ have no influence on the peak current of 2 nM Co2+. This was probably attributed to the selectivity of the complex ligand and the more negative accumulation potential (−1.00 V) which avoid effectively the interference of most metal ions as well as Ni2+. However, the effect of interferences on real samples analysis was eliminated effectively by using the standard addition method.

Fig. 6
figure 6

Effect of possible interfering species on the Co2+ determination at the SnBiE. Solution composition and other conditions are the same as in Fig. 3

Full size image

Real sample analysis

The accuracy of the method was estimated by determining cobalt in real water samples. For the performance of the proposed SnBiE in real sample analysis, the contents of Co2+ in tap water were analyzed by using the standard addition method. Figure 7 shows the AdCSV responses of tap water 2 at the SnBiE. The voltammetric peak current of Co2+ in tap water can be detected at SnBiE, which indicates the possibility of successful determination of Co2+ in real samples by using standard addition method. The concentration of Co2+ in real sample can be calculated from fitted curve (inset of Fig. 7). These results show that the SnBiE exhibits high sensitivity and good selectivity for the determination of Co2+ under the optimum experimental conditions. Additionally, in order to illustrate its accuracy in practical analysis, a comparison was carried out between the SnBiE and ICP-MS for detection of Co2+ in real water samples (Table 3). As can be seen from Table 3, the results obtained from the SnBiE are consistently well with those results detected by ICP-MS, which indicates that the SnBiE can be used for determination of Co2+ in real samples.

Fig. 7
figure 7

Square wave voltammograms obtained in the course of Co2+ determination in tap water 2. The addition concentrations of Co2+ from top to bottom are 0, 2, 4, 6 nM. The inset is the fitted curve of Co2+ measurement obtained from standard addition. Other conditions are the same as in Fig. 3. (When not visible, error bars are smaller than the symbols)

Full size image
Table 3 Comparison of the SnBiE and ICP-MS for determination of Co2+ in real samples
Full size table

Conclusions

In this paper, for the first time, we have presented the suitability of SnBiE for the determination of cobalt in water samples in combination with the AdCSV. Under optimized experimental conditions, excellent linear dynamic range (0.2–20 nM) with a correlation coefficient of 0.998 and a detection limit of 44 pM was obtained in connection with an accumulation time of 60 s. It was found that the performance of the SnBiE for Co2+ analysis was better than that of the mercury, bismuth, lead, and lead–copper electrodes. Real water samples were determined, and the results obtained from the proposed SnBiE were in good agreement with those results detected by ICP-MS. A simple, convenient, and sensitive approach for the determination of Co2+ in natural water samples was provided without the sample pretreatment. Such an application of the SnBiE encourages the on-going efforts for on-site environmental monitoring of cobalt and other heavy metals in place of toxic mercury electrodes.