Introduction

Cobalt is known to be an essential element at trace level to humans, animals and plants for metabolic processes. Cobalt helps to repair the myelin sheath, pernicious anemia, and building of red blood cells. More attention has been focused on the toxicity of cobalt in large concentrations, since it has been found that cobalt can cause vasodilation, flushing and cardiomyopathy in humans and animals. It was reported that cobalt-mediated free radical generation contributes to interfere with DNA repair processes (Valko et al. 2006; Lison et al. 2001; Yaman 2006). The International Agency for Research on Cancer (IARC) has classified this metal as possibly carcinogenic to humans (IARC 1991).

Except occupational sources, main sources of cobalt are foods and beverages. World Health Organization (WHO) recommends that 0.005–0.010 mg of cobalt might be the daily optimal intake for adults (Jiang et al. 2005). Therefore, the determination of Co in water, food, and environmental samples is important in the fields of environmental analysis and medicine.

Co concentrations in water matrices were found in the ranges of <1.5–15.1 ng mL−1 for sea water, 0.5–35.4 ng mL−1 for river water, 0.3–22.3 ng mL−1 for tap water, and 2.0–2.3 ng mL−1 for mineral water.

Due to the allowable low levels of cobalt in food and beverages, and its very low concentrations in natural waters as described above, reliable and sensitive analytical methods are required for determination of this metal. For this purpose, preconcentration techniques in connection with the analytical methods such as electrothermal atomic absorption spectrometry (ETAAS), inductively coupled plasma-atomic emission spectrometry (ICP-AES) and inductively coupled plasma-mass spectrometry (ICP-MS) have been used (Sawula 2004; Guo et al. 2004a; Tanaka et al. 2002). As a result, the use of preconcentration techniques in the determination of Co by flame atomic absorption spectrometry (FAAS) is still important because of the disadvantages such as the interferences from matrix components, the necessity of expert for usage, and the high costs of other methods, relatively (Tuzen et al. 2005; Venkatesh et al. 2005; Guo et al. 2004a; Cadore et al. 2005; Chen and Teo 2001; Yaman and Gucer 1995). Among the preconcentration techniques, solid phase extraction (SPE) has received the most attention due to its simplicity, high concentration factor, and more environmentally friendly reagents used. Moreover, SPE and solvent extraction were also used for speciation studies (Yaman 2003a, b; Yaman and Kaya 2005).

In the case of water analysis, SPE is more advantageous because the large sample volume doesnot cause any problem in preconcentration steps, relatively (Camel 2003; Yaman 2003a, b). Although a great number of studies have been reported using SPE such as ion exchange resins, activated carbon, chelating resins and others, most of them have no practical application due to the necessary preconcentration time as high as 400 min, and typically ranging from 100 to 400 min (Guo et al. 2004b; Narin et al. 2000). Amberlite XAD resins have received an increased attention as basic matrixes for designing new chelating resins. The most successful and popular applications of the Amberlite XAD resins have been carried out using Amberlite XAD-7 resin having a hydrophilic surface and intermediate polarity, as a highly crosslinked macroreticular acrylic resin (polyacrylic acid ester polymer). The higher surface area and sorption capacity of Amberlite XAD-7 resin (Tewari and Singh 2000) than from Amberlite XAD-2 resin (surface area of 450 m2 g−1 for XAD-7, and surface area of 330 m2 g−1 for XAD-2) makes it superior.

Among the commonly used ligands, 4-(2-pyridyl-azo) resorcinol (PAR) is a versatile organic chromogenic reagent and it forms complexes with a variety of transition metals at different pH ranges and adsorbents. The other advantage of PAR is the absence of its affinity for alkali and alkaline-earth ions.

In this study, a preconcentration method based on adsorption of Co on the PAR-loaded-XAD-7 was modified for determination of Co at the nanogram per milliliter level by FAAS. The optimized method was applied for Co determination in different natural mineral waters.

Experimental

Apparatus

An ATI UNICAM Model 929 flame atomic absorption spectrophometer (FAAS) equipped with ATI UNICAM hollow cathode lamp (HCL) was used for determinations (ATI UNICAM, Cambridge UK). The optimum conditions for FAAS are given in Table 1. The pH was measured with (SCHOTT LabStar pH) pH meter. In the preconcentration procedure, magnetic stirrers (Velp scientific, Milano, Italy) and centrifuge were used.

Table 1 Operating parameters for flame atomic absorption spectrophotometry (FAAS)
Full size table

Reagents and solutions

All glass apparatus (Pyrex) were kept permanently full of 1 M nitric acid when not in use. In the digestion work, concentrated HNO3 was used to extract the adsorbed Co from the PAR-loaded XAD-7. The diluted standard solutions were prepared from stock standard cobalt solution of 1000 mg L−1 (Merck).

The buffer solutions in the range of pH 1.5–6.0 ± 0.2 were prepared using 0.1 M citric acid plus 0.1 M HCl/0.1 M NaOH solutions. PAR solution of 0.02% was prepared by dissolving 0.02 g of 4-(2-pyridyl-azo) resorcinol in 100 mL of 0.1 M NaOH. A 1.0 g of Amberlite XAD-7 (surface area, 450 m2 g−1 and bead size, 20–40 mesh) taken from Aldrich (Milwaukee, USA) was pretreated with 1 M NaOH for 1 h, and 4 M HCl for 1 h. The mixture was separated by filtration, washed by distilled water and passed into methanol. The dried XAD-7 resin was ground using agate mortar.

Preparation of PAR-loaded XAD-7

PAR-loaded Amberlite XAD-7 was prepared as follows: 600 mg of pretreated XAD-7 was suspended in 100 mL of 0.02% PAR solution and the pH of this mixture was adjusted to the desired value by adding the dilute solutions of HCl and/or NaOH. Then, the necessary buffer solution (10 mL) was added and the mixture was stirred for 30 min. This mixture was filtered through a filter paper and the residue on filter paper was dried at 60°C.

Preconcentration procedure

The 100 mg of the dried PAR-loaded XAD-7 was added to the Co solution of 60 mL (40 ng mL−1). The cobalt solution contains matrix components in the following levels (mg L−1): Ca 100, Mg 50, Fe, Zn and Al 2, so that it represents natural water. This solution is referred to as the “model solution” in this study. Then, a 10-mL buffer solution was added. The pH of this suspension was adjusted to the studied pH. The mixture was mechanically stirred for 45 min and filtered through a filter paper (Advantec Toyo 5 B, white ribbon). After drying the residue on filter paper at 60°C, the residue was transferred to a glass beaker by scraping. For decomposing the adsorbed Co on the PAR-loaded resin, 4 mL of concentrated HNO3 was added and the mixture was evaporated to near dryness. After cooling, 3 mL of 1.5 M HNO3 was added, mixed, and centrifuged to take the extracted metal into the solution. The clear solution was measured using FAAS. The steps of preconcentration scheme are given in Fig. 1. The blank solutions were carried out in the same way.

Fig. 1
figure 1

Steps of analytical scheme in the enrichment procedure

Full size image

Results and discussion

Chelating resins have been used to preconcentrate metal ions after a chelating agent was adsorbed or bonded chemically on Amberlite XAD resin surface. Although chelating resin bonded chemically is so stable that it can be used repeatedly, its adsorption capacity to metal ions is not high because of the steric effect of the chelating agent. In addition, it is not easy to select a desorption solvent, because the desorption of electrostatically bound metal ions is expected to be achieved only by proton exchange from a concentrated acidic solution. On the other hand, although the adsorbed chelating resin is less stable than the bonded chelating resin, it has more capacity site and is more effective for desorption, because of its ability to utilize either a common organic solvent or an acidic solution to take-off the adsorbed metal chelates or the bound metal ions. Therefore, if an appropriate chelating agent adsorbed stably was selected, the adsorbed chelating resin would be rather favorable for the separation and enrichment of metal ions. Enhancement of the adsorption capacity of chelating resin can be achieved by increasing the number of chelating sites on the resin as well as their accessibility. This can be obtained using a resin which has a relatively high surface area and by selecting a chelating agent of small molecular size. Amberlite XAD-7, a polyacrylic acid ester polymer, has a hydrophilic surface and a relatively high surface area (450 m2 g−1).

The parameters that are thought to affect the preconcentration and measurement steps in the analytical scheme were examined using the model solutions of 40 ng mL−1. The effect of each parameter was tested three times.

The influence of pH on recovery

Hydrogen-ion concentration plays an important role in the preconcentration of metals by SPE because it significantly affects the metal-ligand complex formation.

The model Co solutions were preconcentrated as shown in Fig. 1 using various pH values ranging from 1.5 to 6.0. The obtained recoveries are given in Fig. 2. It was seen that the maximum recoveries (up to 90%) were found in the pH range of 1.5–2.5 for both first and second steps. Therefore, for subsequent studies, the pH 2.0 ± 0.2 was used for both the two steps.

Fig. 2
figure 2

The influence of pH on recovery of Co with PAR-loaded XAD-7

Full size image

Effect of the amount of PAR on recovery

In order to determine the optimum amount of PAR, the recoveries of cobalt at the optimum pH (2.0 ± 0.2) were examined using different volumes of 0.02% PAR by adding 60 mg of pretreated Amberlite XAD-7. It is seen from Fig. 3 that the recoveries increase up to 90% with 15-mL PAR and do not change by adding up to 30 mL. Thus, 15 mL of 0.02% PAR was used for 60 mg of XAD-7, in the subsequent studies. This amount was increased according to the amount of XAD-7, proportionally.

Fig. 3
figure 3

The influence of amount of PAR on recovery of Co

Full size image

Effect of the amount of the PAR-loaded Amberlite XAD-7 on recovery

In order to determine the optimum amount of the adsorbent resin, different amounts of the PAR-loaded XAD-7 were added to the model solutions described above.

The preconcentration procedure with all other optimum conditions (pH 2.0 and 15 mL PAR solution) was applied for preconcentration. It was found from Fig. 4 that the recoveries increased up to 90% by adding 100 mg PAR-loaded XAD-7 and did not change by increasing up to 150 mg. Thus, 100 mg of the PAR-loaded XAD-7 was used for 60 mL of the sample solution, in all subsequent studies. This amount was increased in proportion to the sample volume.

Fig. 4
figure 4

Determination of optimum amount of PAR-loaded XAD-7

Full size image

Effect of the stirring time on recovery

The kinetics of the Co(II) sorption and desorption were investigated in a batch system. It was found that the stirring time of 30 min is sufficient for first step in Fig. 1. The preconcentration procedure was applied to the model solutions using different stirring times with all other optimum conditions. It was found that the stirring time of 45 min was sufficient for maximum recovery in the second step (up to 90%). This stirring time is shorter compared to other studies where preconcentration time as long as 2–4 h was required for maximum recovery in the literature (Chakrapani et al. 1998). So, the stirring time of 45 min was used for all studies.

Effect of the elution volume on recovery

After decomposition by concentrated nitric acid as described above, the efficiency of the eluent (1.5 M HNO3) was studied by taking different volumes (1.5–4.0 mL). It was found that 3 mL of 1.5 M HNO3 was sufficient for maximum recovery (90%) of Co (Fig. 5). Therefore, 3.0 mL of 1.5 M HNO3 was used for complete desorption of the cobalt.

Fig. 5
figure 5

Effect of the final volume of the eluant on recovery

Full size image

Effect of the buffer volume on recovery

Test solutions were buffered in order to have stable pH values and avoid time-consuming dropwise addition of acid or base to obtain the desired pH. The influence of buffer amount was carried out by keeping all other experimental variables constant. The results showed that by adding above 5.0 mL of buffer solution to 60 mL of solution, no obvious variation takes place in the extraction yield for both first and second steps (Fig. 6). Thus, a 7.0 ± 2-mL aliquot of buffer solution was added for all subsequent experiments and this volume was increased in proportion to the sample volume.

Fig. 6
figure 6

Effect of the amount of buffer solution on recovery

Full size image

Analytical performance

The calibration curve was observed to be linear at the concentration range of 0.1–1.5 mg L−1 using direct FAAS. Therefore, the preconcentration factor of 200-times was required due to cobalt concentration as low as 0.5 ng mL−1 in mineral waters. The calibration curve was observed to be linear at the concentration range of 0.5–5.0 ng mL−1 by taking 600 mL solution to final volume of 3.0 mL of 1.5 M HNO3 (using 1000 mg PAR-loaded XAD-7 and 50 ml of buffer solution). Although the recovery was found to be 87% using 600 ml of model solution, the calibration curve was obtained rectilinear due to the precise recoveries. Therefore, the preconcentration factor of 200-times was achieved. The equations of the curves are as follows:

$$ Y = 76X - 0.7\quad R^{2} = 0.9997\quad \quad {\text{for}}\,\,0.1 - 1.5\,{\text{mg}}/{\text{L}}\,{\text{by}}\,{\text{FAAS}} $$
$$ Y = 13.4X - 0.3\quad R^{2} = 0.9995\quad \quad {\text{for }}0.5 - 5.0\,{\text{ng}}/{\text{mL by preconcentration-FAAS}} $$

Relative standard deviations (RSD) were 13% for 600 mL of 2.0 ng mL−1, for 10 replicate preconcentration procedures. The level of Co in the blank was found to be 0.1 ng mL−1 with a standard deviation of 0.033 ng mL−1. Therefore, the detection limit (LOD) defined as three times the standard deviation of the blank was 0.1 ng mL−1 when 600 mL of solution was preconcentrated to a final volume of 3.0 mL.

The accuracy of the method was studied by examining the Standard Reference Material; SPS-SW2 Batch 113 surface waters (Spectrapure Standards AS-oslo, Norway). The results are given in Table 2. It can be seen that the recovery value was found to be 98% for Co. This can be attributed to use the obtained calibration curve at the same conditions. In addition, the accuracy of the method was studied by examining the recoveries of cobalt from mineral water samples fortified with this element. From the results of Table 2, it was found that, at least, 90% of the cobalt added to water samples was recovered. This can be attributed to the losses in preconcentration steps.

Table 2 Co concentrations determined in natural mineral waters in this study
Full size table

Applications

The optimized preconcentration method was applied to the determination of cobalt in various natural mineral waters taken from the markets in Elazig city, Turkey. According to the concentration of Co ions in the studied samples, a 600 mL of water samples was transferred into a beaker. The optimized preconcentration method was applied to the determination of Co in these samples. The obtained results are given in Table 2. The given values are the mean values of three different portions of the same sample. The Co concentrations were found to be in the ranges of 0.5–3.5 ng mL−1. These concentrations are much lower than the allowed level. Soylak et al. (2004) reported the cobalt concentration in bottled mineral water as 5.5 ng ml−1.

Conclusion

A sensitive, selective and reliable preconcentration method was modified by adsorption of cobalt on the PAR-loaded Amberlite XAD-7 for determination of Co at the ultratrace level. Matrix components of mineral water samples were characterized and mean concentration levels of matrix components were added to all standard Co solutions. The sensitivity of FAAS was increased up to 200-times using the optimized method. So, the detection limit was found to be 0.1 ng mL−1 when a sample volume of 600 mL was preconcentrated to a final volume of 3.0 mL. The obtained optimum pH at low value (pH 2) presents high advantage because the major and minor compounds can precipitate and interfere with the measurements at high pH values.