Multivariate optimization of differential pulse polarographic–catalytic hydrogen wave technique for the determination of nickel(II) in real samples

2.1 Apparatus

A Metrohm model 797 VA Computrace three electrode system (Metrohm Herisau, Switzerland) consists of the hanging mercury dropping electrode (HMDE) as the working electrode for cyclic voltammetric measurements, while the dropping mercury electrode (DME) was used as the working electrode to obtain all the polarographic waves. The reference electrode was Ag/AgCl and platinum wire was used as the counter electrode. All pH measurements were carried out with Crison micro pH model 2000 pH meter. All grape samples were digested with a temperature controllable 100 Teflon PFA vessel microwave digester (CEM MARS-5, CEM Co., USA).

2.2 Chemicals and reagents

All the experiments were performed at 25 °C using freshly prepared solutions. Double distilled mercury was supplied by Metrohm (Durban, South Africa (SA)) and deionized water was generated from an aqua MAX™ – Basic 360 series water purification system from TRILAB SUPPORT (Durban, SA). The dissolved oxygen in the solutions was removed by passing 99.8% pure nitrogen gas (AFROX Durban, SA) for 15 min. Nickel(II) standard solution was prepared by accurately weighing 4.050 g L⁻¹ of NiCl₂. 6H₂O (Merck Laboratory Suppliers Pty Ltd, RSA) to get approximately 1.0 μg L⁻¹ and by adding 3.0 mL of concentrated HNO₃ corresponding to the anions of the salts to suppress the hydrolysis. 1.0 M ammonium chloride was prepared by weighing 53.49 g and dissolving in 1.0 L with deionised water and the pH adjusted with 5% (v/v) ammonium hydroxide and 1% (v/v) hydrochloric acid solutions. Carbon disulphide, piperidine and morpholine were also purchased from Merck Laboratory Suppliers Pty Ltd, SA.

2.2.1

2.2.1 Synthesis of ammonium piperidine/morpholine dithiocarbamates (APDC/AMDC)

Carbon disulphide (80 g) was slowly added to each solution of (85 g) piperidine and morpholine dissolved to 25 mL with distilled water at 5 °C under constant stirring, followed by addition of 20% (v/v) ammonium hydroxide for neutralization. The resultant reaction mixture was warmed at room temperature and washed repeatedly with acetone of 99.5% purity. The product was purified by recrystallization in acetone (Hemasundaram, 2006) with melting points ranging from 196–199 °C and 182–185 °C for APDC and AMDC respectively, at 740 mm pressure as shown in Scheme 1.

Close

Scheme 1

Reaction route for the preparation of ammonium salt of piperidine/morpholine dithiocarbamates.

Export to PPT

2.3 DPP-CHW methodology

A measured volume of the NH₄Cl-NH₄OH buffer and ligands (APDC/AMDC) was added to adequate amounts of electroactive species [nickel(II)]. Dissolved oxygen was removed by bubbling pure nitrogen through the solution in the polarographic cell for 15 min. Polarograms of the solutions were recorded using differential pulse polarography (DPP) at −1.22 V vs Ag/AgCl in ammonium chloride-ammonium hydroxide medium for APDC and AMDC respectively. All cyclic voltammograms were recorded in HMDE mode.

2.4 Preparation of grape samples

A total of 92 grape samples (46 red and 46 white) were collected from local markets and also directly from the farm houses. The grapes were rinsed with 20% (v/v) HNO₃ and stored at 4 °C in plastic bags. Six of each type of grape samples were used for recovery studies and the rest (10 red and 10 white from one source) were subdivided into two sets. The first set containing 5 red and 5 white was washed manually for 5 min with deionized water while the other set was analysed without any washing. The investigation was done in epicarp (grape peel), endocarp (grape pulp) with seeds and whole grapes giving a total of 120 samples excluding the spiked samples.

The three sample sets were packed in a decontaminated poly vinyl chloride tube and placed in a micro-wave oven separately at 60 °C for 10 h for epicarp, 12 h for endocarp along with seeds, and 15 h for the whole grape samples. The dried samples were powdered and transferred into a closed 50 mL Teflon container for digestion process. Approximately 1.0 g of powdered sample was accurately weighed and transferred into a closed Teflon container, to which 2.0 mL of HNO₃, 1.0 mL of HCl and 0.5 mL of H₂O₂ were added. The whole sample system was placed in a micro-wave oven at 90 °C for 2 h to complete the digestion process then transferred into a decontaminated poly vinyl chloride tube and diluted with deionized water up to 10 mL and used as a sample for nickel(II) analysis as mentioned above.

2.5 Experimental design

A Box–Behnken design with fifteen runs, three independent variables (pH, scan rate and concentration of ligand) and three replicates at a centre point was used in this study. The experiment was designed such that it was randomized to reduce confounding variables by equalizing the three independent variables that have not been accounted for in the experimental design. Table 1 shows the experimental design of the three factors pH, concentration of dithiocarbamates (DCs) and scan rate. The maximum current signals and their respective potentials due to nickel(II) were taken as the response of the designed experiments per DC for the three replicates. Since the two ligands (APDC/AMDC) were used in the study, the design was therefore duplicated with experiment 1 referring to the analysis performed in the presence of APDC, while experiment 2 refers to AMDC. The statistical analysis on the response of the model was performed with an analysis of variance (ANOVA), pareto charts, interaction plots and surface responses. Interestingly, the design enabled us to evaluate the responses with respect to the simultaneous factor variations in all the experimental regions studied and to optimize the experimental conditions for the detection of nickel(II), taking into account randomization for even distribution, as illustrated in Fig. 1.

Close

Figure 1

Shows the random distribution of the three factors pH (X1), concentration of DCs (X2) and scan rate (X3).

Export to PPT

2.6 Data analysis

STATGRAPHICS Plus version 5.1 and Microsoft excel® 2010 were used for data evaluation and preparation of the experimental design. Peak evaluation was performed with 797 PC Software 1.3® 2008.

3.1 Analysis of variance on analyte responses

The standardized pareto charts in Fig. 2a–b show the estimates in a decreasing order of importance. The ANOVA results reported in Table 2 describe the variability in current response into separate pieces for each of the effects. The statistical significance of each effect shown in the pareto charts was tested by comparing the mean square deviation against an estimate of the experimental error. With regard to experiment 1 (APDC), only one effect (scan rate) had a P-value <0.05, and the 85.59% of the variability in analyte was explained by the model based on the R-squared value. In the case of experiment 2 (AMDC), three effects (pH, concentration of ligands, scan rate) A: pH, AA and AB had P-values of 0.0028, 0.0225 and 0.0419 respectively, which were <0.05. This is a clear indication that they are significantly different from zero at the 95.0% confidence level, while the R-squared statistic indicated that the model demonstrated a 90.89% of the variability in nickel(II) responses. Furthermore, a multifactor analysis of variance for the current response was performed to test significant interactions amongst the factors, using sufficient data obtained through the Box–Beckhen experimental design. With regard to experiment 1, multifactor analysis of variance showed that all the tested factors were statistically insignificant at the 90.0% confidence level, with pH having the lowest P-values close to 0.05. The highest P value was 0.546 belonging to the concentration of AMDC, while in experiment 2 it was 0.856 from the scan rate. Since the P-value in both the experiments was greater than 0.05, there was no indication of serial autocorrelation in the residuals. Further analyses of the extent of the confounding variables amongst the effects were evaluated by the correlation matrix of the three pairs. For the highest scan rate and pH, the peaks were much broader and their potential was more positive as shown in Table 1.

Close

Figure 2

Shows the standardized pareto charts for current response using (a) APDC and (b) AMDC.

Export to PPT

3.2 Analysis of Box–Behnken design

For each factor, the zone of the experimental domain corresponding to a high value of peak heights deduced from the response surfaces was selected as the proper experimental domain. Thus the interaction plots for the nickel(II) current signals presented in Fig. 3a–b, were based on (+) and (−) values showing the maximum and minimum values of each parameter respectively (Rioa et al., 2005). These interaction plots also showed the estimated variable as a function of pairs of factors. In Fig. 3b, it can be seen that the important interaction exists between pH and concentration of AMDC (positive), which is more important than the effects of pH alone. This is highlighted in Fig. 3d, where the lowering of the pH leads to a better current response. However, when the same experiment was conducted with the APDC, a marginable increase in current was observed at a lower pH, as shown in Fig. 3a. This result suggests the dependency of the mechanism of detection of the protons from the ligands, due to the zero valent metal complex at the electrode. Surface plots shown in Fig. 3c–d represent the simultaneous effect of pH, concentration of the ligand and scan rate on the current signal by nickel(II). The scan rate was kept constant at 3.0 mV/s and 8.0 mV/s for experiments 1 and 2 respectively. Very little effect of the concentration of APDC on the nickel(II) current signal was observed (Fig. 3) within the selected range, however with a decrease in pH, nickel(II) current signal improved to the highest value of approximately 2400 nA at pH 3.0 for a 1.0 ppm standard. Furthermore, the models predicted that the path of steepest decent for the analyte response was from pH 6.0 onwards. This is the path from the centre of the experimental region along which the estimated response changes rapidly for the smallest change in the experimental factors. In both the experiments, there are six points that were generated as the path of steepest descent by changing the pH in increments of 1.0 (results not shown). Beyond the main effects on current response by each factor, the experimental design further revealed that there was a significant interaction between pH and concentration, and between pH and scan rate. These interactions were more predominant in experiment 1 as can be seen in the pareto charts shown in Fig. 2. Further analysis showing the extent of the confounding variables is shown in Fig. 4a–b. In these models there are three pairs of effects shown in the interaction diagrams, and they show non-zero correlations. However, we decided to focus on the concentration of the DCs and pH, since they influenced prominently both experiments as can be seen in the pareto chart in Fig. 2. Both the contours of the surface responses showed a diagonal matrix with 1’s on the diagonal and 0’s off the diagonal for pH around 7.0. The distance between the diagonal lines tends to increase as the pH increases; this justified the low optimum values of the pH obtained.

Close

Figure 3

(a) and (b) Interaction plots for current response at different combinations of the three variables. (c) and (d) show the estimated response surfaces for experiment 1 and 2 respectively.

Export to PPT

Close

Figure 4

(a) and (b) Show contours of estimated response surface, a clear indication of the relationships between pH (X1) and concentration of DCs (X2) at optimum scan rates (X3).

Export to PPT

3.3 Electrochemical studies

Scheme 1 shows the structures of the APDC/AMDC considered in this study. They form co-ordination with the sulphur atoms of DCs, being noteworthy for their remarkable versatility as a strong chelating agent. In the DPP-CHW methodology, nickel(II) forms a complex with APDC/AMDC, which subsequently absorbs onto the DME working electrode during the deposition step. During the polarographic scan, nickel(II)-Dithiocarbamate [Ni⁺² -DC]_ads complex is reduced to [Ni⁰ -DC]_ads which undergoes protonation by accepting proton from the solution [Ni⁰ –DCH⁺]_ads in the adsorbed state, followed by the reduction to liberate hydrogen in the presence of mercury electrode, as shown in Eqs. 1–3 below:

Close

Figure 5

Cyclic voltammograms for (a) 5.0 mM APDC in 0.4 M NH₄Cl at pH 3.0, (b) 1.0 mM AMDC in 0.3 M NH₄Cl at pH 3.0, (c) 5.0 mM APDC + 0.05 ppm nickel(II) in 0.4 M NH₄Cl at pH 3.0, (d) 1.0 mM AMDC + 0.05 ppm nickel(II) in 0.3 M NH₄Cl at pH 3.0.

Export to PPT

Therefore, oxidation of DCs at the HMDE is represented as $$\text{RSH}+\text{Hg}\to \text{RSHg}+{\text{e}}^{-}+{\text{H}}^{+}$$ $$\text{RSH}=\text{Dithiocarbamate}$$ in agreement with the earlier reports in the literature (Stricks and Chakravarthi, 1962). The typical polarograms were recorded at E_1/2 = −1.22 V for both APDC/AMDC-nickel(II) complex systems after optimization studies using DPP-CHW as shown in Fig. 6a–b.

Close

Figure 6

Obtained polarograms in grape samples after optimization studies for (a) APDC-nickel(II) and (b) AMDC-nickel(II) system with DPP-CHW.

Export to PPT

3.4 Computational studies

The electron-donating tendencies of piperidine (Fig. 7a) and morpholine (Fig. 7b) adducts with carbon disulphide (CS₂) were monitored by performing their density functional theory (DFT) calculations using the computer Gaussian program (Frisch et al., 2003). The initial structures of both adducts were geometrically optimized at the B3LYP (DFT) level using 6–31 g^∗∗ and 6–311 g∗∗ basis sets. The global minimum for each optimized geometry was further confirmed by a frequency calculation. The energy differences (ΔE) between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) are considered to be an important parameter which defines the chemical activity of any compound, with the smaller value indicating a stronger tendency to donate electrons. Accordingly, the HOMO and LUMO of both piperidine and morpholine adducts were generated, and the energy data are summarized in Table 3. The HOMO–LUMO of structures, obtained using 6–31 g^∗∗ basis sets, are depicted in Fig. 8a–d. Generally, the distribution of electrons in HOMO and LUMO of both complexes was almost similar (Fig. 8a–d). As expected, the electron density of both piperidine (Fig. 8a) and morpholine (Fig. 8c) adducts was mostly localized on the sulphur atoms, suggesting a stronger propensity of these atoms to participate in the complexation with the electron deficient species including metals. However in case of piperidine, the comparatively lower energy gaps (ΔE) between HOMO and LUMO shown in Table 3, suggest a better tendency to act as an electron donor and support the DPP studies, where the intensity of current was observed to be higher in case of piperidine-nickel(II) complex, than with the morpholine-nickel(II) complex. It is believed that the negative inductive effect (-I effect) of the oxygen atom in the morpholine adduct, reduces the electron density on its nitrogen atom through the intervening electrons of the bonds, and probably accounts for the higher ΔE values shown in Table 3. Based on these results, it is believed that the energy barrier between HOMO and LUMO of ligands could be a very useful parameter to predict their binding affinities in metal complexation, and could therefore play a pivotal role in identifying the more efficient and sensitive ligands in the field of electroanalytical chemistry.

Close

Figure 7

Two dimensional representations of (a) Piperidine and (b) Morpholine adducts with CS₂.

Export to PPT

Close

Figure 8

Electron density of (a) HOMO and (b) LUMO for piperidine and electron density of (c) HOMO and (d) LUMO for morpholine, adducts with CS₂, computed at b3lyp/6–31 g^∗∗ energy level.

Export to PPT

3.5 Analytical applications

The proposed DPP-CHW technique was applied for the analysis of nickel(II) in different grape samples, using the optimum parameters (shown in Table 4) obtained through the Box Behnken experimental design. Different grape samples were collected from local markets and directly from the farm houses. Samples were treated as described in the experimental section before the quantification of nickel(II). The obtained data are summarized in Tables 5–8.

In this study, 92 different forms of grape (red and white) samples were analysed for the determination of nickel(II). Significant differences of nickel(II) concentrations were observed between the epicarp, endocarp and the whole grape samples collected from local markets and farm houses. On the whole, no significant differences between the samples (epicarp, endocarp and whole grape) were observed. Table 8, clearly shows that the recoveries of nickel(II) were higher with APDC than with AMDC, which was indicative of the sensitivity of APDC. This study was carried out with three different concentrations ranging from 10.0 to 20.0 μg L⁻¹. Overall, for both ligands (APDC/AMDC), recoveries obtained were always higher than 97.0% as shown in Table 8.

3.6 Comparison of the DPP-CHW method with reported methods

Table 9 shows the comparison of multivariate optimization voltammetric determination of nickel(II) in real samples with other methods reported in the literature. The LOQ’s for APDC & AMDC are 2.7 × 10⁻⁶ and 1.8 × 10⁻⁶ M respectively for the determination of nickel(II) in different grape samples, that are better than those reported in the literature (Dadfarnia et al., 2010). Additionally, the APDC/AMDC could be considered as more promising and also cost effective chelating agents for the analysis of nickel(II), in contrast to those reported in the literature [C4MIM][PF6] (Dadfarnia et al., 2010) MSOPD (Khorrami et al., 2004) and Py-MCM-48 (Sadeghi et al., 2011). Although previously reported chelating agents such as EDTA (Baytak and Turker, 2006), MSOPD (Khorrami et al., 2004), PAN (Bidabadi et al., 2009) and Py-MCM-48 (Sadeghi et al., 2011) are sensitive than the current APDC/AMDC, the methods employed required more solvent (SPE, LLE) with an elaborate sampling procedure utilizing sophisticated instrumentation such as FI-FAAS, ETAAS and GFAAS. Accordingly, the present method for the determination of nickel(II) in real samples therefore is facile, sensitive, economical and environmentally friendly.

3.7 Effect of foreign ions

In the present study, various heavy metal ions commonly associated with nickel(II) such as aluminium(III), chromium(VI), zinc(II), molybdenum(VI), selenium(IV), tellurium(IV), cerium(IV) and tin(II) have been evaluated for their impact in the DPP-CHW of dithiocarbamate-nickel(II) system. In particular, the concentration of nickel(II) was maintained at 4.0 ppm with a 100 fold excess of foreign ions added. Among the metal ions studied, iron(II) and coper(II) formed precipitates with DCs in the conditions where nickel(II)-dithiocarbamate complex yielded a CHW. Aluminium(III) did not interfere whereas chromium(VI) gave a CHW with peak potentials at – 1.56 V and – 1.65 V vs Ag/AgCl, suggesting that a simultaneous determination of nickel(II) & chromium(VI) was feasible without any separation or adding masking agents. Zinc(II) interfered in the determination of nickel(II), but were masked by adding 5.0 mL of 2% sodium tartrate solution. Molybdenum(VI) severely interfered by increasing the peak height of nickel(II) and shifting the peak potentials towards a more negative value. The selenium(IV), tellurium(IV), cerium(IV) and tin(II) also did not interfere with the nickel(II)-dithiocarbamate system. Anions such as fluoride, bromide, iodide, tartrate, sulphate, thiosulphate, phosphate, carbonate, oxalate, nitrite and nitrate did not interfere with the CHW of dithiocarbamates-nickel(II) system. Perchlorate and thiocyanate interfered with the wave by reducing the wave height by approximately 40.0% and EDTA interferes severely by completely suppressing the nickel(II) catalytic hydrogen wave.