Solid-phase extraction of vanadium(V) from tea infusions and wines on immobilized nanometer titanium dioxide followed by ICP-OES analysis

2.1 Chemicals and instruments

HCl 37%, VOSO₄ · xH₂O, titanium (IV) tetraisopropoxide (97% v/v), 2-propanol (99.5% v/v) and vanadium stock solution 1.02 g L⁻¹ were supplied from Sigma–Aldrich, NH₄VO₃ from Riedel-De Haën Ag Seelze-Hannover and silica 63 μm from Merck. 200 mg L⁻¹ V(V) standard solution was prepared by dissolving NH₄VO₃ in deionized water while V(IV) stock solution was prepared from VOSO₄ · xH₂O in 0.2 M HClO₄. The exact vanadium concentrations of both solutions were determined by ICP-OES. Ultra-pure water (resistivity 18.2 MΩ cm⁻¹ at 25 °C) was produced in laboratory by means of a Millipore Milli-Q system.

An ICP-OES Perkin Elmer Optima 3300 DV was used for all the measurements, following the operating conditions suggested by the manufacturer (plasma flow 15 L min⁻¹, aux flow 0.5 L min⁻¹, neb flow 0.85 L min⁻¹, RF power 1300 W, sample flow rate 1.5 mL min⁻¹, λ 292.402 nm). A four-point calibration curve, mean of three independent calibration lines, was generated in the concentration range 0.01–10.2 mg L⁻¹. Instrumental detection limit (IDL) and instrumental quantification limit (IQL) were calculated as 3 and 10 times the standard deviation related to the blank signal, respectively. IDL was 3 μg L⁻¹ and IQL was 10 μg L⁻¹.

A Shimadzu AA-6601G/GFA 6500 ET-AAS system was used for total vanadium determination (Teissèdre et al., 1998).

A Philips PW 1830 generator and a Cambridge S360 microscope were employed for XRPD and SEM analysis. ASAP 2010 physisorption analyzer (Micromeritics Instrument Corp) was used for BET and porosity analysis; samples were pre-treated at 200 °C in vacuum before measurement.

2.2 Synthesis of nanosized titanium dioxide immobilized on silica gel

Nanosized titanium dioxide immobilized on silica gel (TiO₂/SiO₂) has been prepared by sol–gel method (Caratto et al., 2012; Hench and West, 1990). In the first step, titanium isopropoxide, 2-propanol and ultrapure water were mixed under vigorous stirring for 4 h obtaining a white mellow gel. In the second step, silica gel was immersed in TiO₂-precursor sol and mixed by magnetic stirring or sonication. The coated silica gel was washed two times with ultrapure water in order to remove dispersed TiO₂. Finally the material was dried at 100 °C for 1 h and subsequently calcined at different temperatures in order to obtain crystalline anatase structure.

Three different samples were investigated in this study, ST1 = TiO₂/SiO₂ synthesized under magnetic stirring and calcined at 450 °C for 1 h; ST2 = TiO₂/SiO₂ synthesized under sonication and calcined at 450 °C for 1 h; ST3 = TiO₂/SiO₂ synthesized under magnetic stirring and calcined at 550 °C for 1 h.

2.3 Column preparation

Polypropylene columns (1.5–4 mL, Alltech, Italy) were filled with 0.05–1 g of nanosized TiO₂ immobilized on silica gel and conditioned to the native sample pH with 20 mL of 0.1 mol L⁻¹ CH₃COONa buffer solution.

2.4 Sample preparation

For the optimization of the procedure, two types of wines (homemade white wine and Barbera from Oltrepò Pavese, Pavia) and a commercial no brand tea infusion (prepared by immersion of one tea bag in 200 mL of hot water for 5 min) were selected and analyzed with no further treatment. The native total vanadium content was measured by atomic absorption spectroscopy with electrothermal atomization (ET-AAS) after acidic digestion (Teissèdre et al., 1998), and V(V) content was determined according to a validated method based on Chelex100 SPE followed by ICP-OES analysis (Sturini et al., 2010). Samples with no modification of the native pH were pre-concentrated (see Section 2.5); in detail, tea infusions and white wine were submitted to SPE as purchased, while red wine was 1:1 diluted.

2.5 Pre-concentration procedure

Samples (25 mL for wines, 50 mL for tea infusions) were delivered at 1 mL min⁻¹ flow rate throughout the columns. The pre-concentration system consisted of a multi-channel peristaltic pump (Gilson, Italy) equipped with a set of Tygon pump tubes to connect the samples with the columns. V(V) retained on the column (1 g sorbent) was quantitatively eluted by 0.1 M HCl solution (0.1 mL min⁻¹) and the effluent fractions were collected and analyzed by ICP-OES (3 × 1 mL for tea infusion, 2 × 1 mL for white wine and 5 × 1 mL for red wine).

3.1 Characterization of nanosized titanium dioxide immobilized on silica gel

The obtained materials were characterized by X-ray powder diffraction (XRPD) technique, scanning electron microscopy (SEM) and BET analysis (Table 1). As shown in Fig. 1, all samples crystallized in anatase structure. It is well known that specific surface area of nanoparticles is related to their size, therefore three different procedures to immobilize nanosized titanium dioxide on silica gel were tested. Stirring and temperature of calcination affect the surface area; the larger one was observed on ST1 sample. The SEM analysis, reported in Fig. 2, showed that ST3 nanoparticles, calcined at 550 °C, were larger than ST1 and ST2, calcined at 450 °C. The grains of ST1 and ST2 maintained regular size and round in shape. Sample ST1 was more homogeneous than ST2 because the magnetic stirring is more controlled than sonication, and also than ST3 because its smaller anatase grains size allowed a more uniform coating.

Close

Figure 1 XRPD spectra of samples ST1 (blue) and ST3 (red) showing crystalline anatase phase. The spectra of ST1 and ST2 are overlapped because of the same temperature and time of calcination.

Export to PPT

Close

Figure 2 SEM images (dimension: 50–200 μm, magnification: 5000–100, EHT: 20.0 kV). Top: cluster of ST1. Middle: cluster of ST2. Bottom: cluster of ST3.

Export to PPT

3.2 Nanosized titanium dioxide immobilized on silica gel as SPE sorbent: explorative experiments

The affinity of V(IV) and V(V) for the three types of immobilized nanometer TiO₂, namely ST1, ST2, ST3, was investigated. 20 mL of V(IV) and V(V) aqueous solutions (1 mg L⁻¹) at different pHs was separately passed through the packed columns (0.05 g) following the procedure above described. For all the sorbents tested, a quantitative adsorption of V(V) (>95%) was observed in the pH range 2.3–7.0, while that of V(IV) was negligible (<5%) at acidic pHs; note that vanadyl cation is stable only at pH < 2.4 because of its easy oxidation to vanadate by dissolved oxygen (Sturini et al., 2010). In the pH range 2.5–7 VO₂⁺ and H₂VO₄⁻/HVO₄²⁻ are the prevalent species of vanadium(V) (Pyrzynska and Wierzbicki, 2004); being the sorbent point of zero charge (PZC) 5.9 ± 0.1, electrostatic interactions play an important role in the adsorption process.

The V(V) pre-concentration procedure was first developed in tap water samples enriched with 250 μg L⁻¹ V(V), and eluent composition and volume, loading and elution flow rates were investigated.

For elution of V(V), aqueous HCl, HNO₃, NaOH and NH₃ at different concentrations (0.1–1 M) were tested. 0.1 M HCl resulted the best eluting solvent, providing a quantitative desorption of vanadium (95%, n = 3) with no interferences in the ICP-OES analysis, while recoveries not higher than 60% were obtained employing either different concentrations or elution solvents. The optimum elution volume was found by collecting and analyzing fractions of 1 mL; 3 × 1 mL eluting solutions were sufficient to entirely elute the analyte from ST1 and ST2, providing at the same time suitable enrichment factors for analytical application. On the contrary, even using higher HCl concentration (1 M) and/or larger elution volumes (up to 8 mL) a recovery not higher than 60% (n = 3, RSD 5%) was found on ST3.

Since the loading flow rate (0.1–2 mL min⁻¹) did not affect the pre-concentration efficiency (data not shown), for practical reasons it was performed at 1 mL min⁻¹. Contrariwise, the elution flow rate significantly affected recovery. Decreasing the flow rate from 1 to 0.1 mL min⁻¹, recovery gradually increased up to 96% (RSD 5%, n = 3) due to a more efficient desorption of the adsorbed analyte.

On the basis of the analytical features above reported and of the SEM, XRPD and BET characterizations, ST1 was preferred for V(V) pre-concentration.

3.3 Analytical evaluation of the method

A series of experiments were first performed to evaluate the V(V) pre-concentration efficiency of ST1 from tap water and from more complex matrices (i.e. wines and tea infusion), and to characterize the main figures of merit of the proposed method.

In order to evaluate the maximum applicable sample volume, 20, 50, 100, and 250 mL of aqueous solutions containing 10 μg of V(V) were loaded on the column (0.05 g ST1) under the conditions above reported. A quantitative retention of V(V) was observed up to 70 mL. 25 mL of solutions at different V(V) concentrations (3–60 mg L⁻¹) was processed on ST1 to determine the adsorption capacity that was found to be 11 mg g⁻¹. The column was proved to be reusable, after reconditioning at sample pH, for at least 10 adsorption-elution cycles, with no significative decrease in the recovery of V(V) (p = 0.05).

To test recovery and precision of the proposed pre-concentration procedure at lower concentrations, additional experiments were carried out on tap water spiked with 1 μg L⁻¹ V(V), obtaining quantitative recovery rates, with mean values up to 92% (see Table 2).

With the aim of developing a straightforward procedure for the determination of V(V) in beverages, further experiments were devoted to evaluate the efficiency of the ST1 sorbent phase in the pre-concentration of V(V) from more complex matrices. A commercial tea infusion and two wine samples were analyzed for vanadium(V) and total vanadium content by two independent validated methods (Chelex 100 SPE-ICP-OES and ET-AAS, respectively, see Section 2.4), and the measured amounts are included in Table 2.

In the absence of a certified reference material for V(V), the accuracy was tested by recovery tests on the same samples enriched with known amounts of V(V), as reported in Table 2. Working in the conditions above reported (0.05 g ST1, loading flow rate 1 mL min⁻¹, elution flow rate 0.1 mL min⁻¹, sample volume 25 mL) the recovery obtained by processing tea infusion and wine samples enriched with 250 μg L⁻¹ V(V), did not exceed 25%. This was due to the partial adsorption of the metal ion on the sorbent phase, as proved by the consistent analyte amounts found in the percolated samples. These findings suggested that two different processes could have hindered V(V) adsorption, namely competition between the metal ion and the matrix constituents for the sorption sites, and/or the formation of stable complexes between V(V) and the organic species present in wine and tea.

To understand the reasons at the basis of these hypotheses, 50 mL of tea infusion with a negligible vanadium content (<1 μg L⁻¹) was passed through the column followed by 5 mL of an aqueous solution containing V(V) (250 μg L⁻¹). In this case, the V(V) not retained was 58%, proving that a competition between the metal ion and the matrix constituents effectively took place. A similar behavior was found for red and white wines. To improve V(V) adsorption from such complex matrices, the sorbent amount was varied from 0.05 to 1.0 g. SPE of 25–50 mL samples spiked with 250 μg L⁻¹ V(V) was performed. Sample volumes of 25 and 50 mL for wines and tea infusions, respectively, provided recoveries up to 98% (n = 3, RSD 5%). The elution was carried out with different volumes of 0.1 M HCl depending on the matrix, that is 3 × 1 mL for tea, 2 × 1 mL for white wine and 5 × 1 mL for red wine. Under these conditions, tea infusion and wine samples spiked at concentrations near to those found in such beverages (Teissèdre et al., 1998) were pre-concentrated obtaining good recovery and intraday precision, as reported in Table 2. The within-laboratory inter-day precision, evaluated on three samples of each beverage independently spiked at the lowest concentration levels and pre-concentrated in different days, provided RSDs < 12%. The method detection and quantification limits (MDLs, MQLs, respectively), reported in Table 2, prove that the proposed procedure is suitable for the determination of V(V) at the concentration levels usually present in actual beverages (Teissèdre et al., 1998).

As it is evident, enrichment factor depends on the sample type. In particular, for such complex matrices rich in polyphenols, flavonoids, organic acids, non-volatile alcohols, sugars, metal ions, anions (Grindlay et al., 2011), the loading volume is obviously lower than that applicable for water. The major coexisting cations and anions present in the actual beverages were measured and reported in Table 3. It can be seen that the presence of these coexisting ions has no influence on the V(V) recovery under the selected conditions (see Table 2). The SPE procedure on nanometer TiO₂/SiO₂ allowed at the same time to perform a cleanup of the sample, thus permitting the final quantification of the analyte in the eluate, with no need for sample pretreatment (e.g. acidic digestion followed by V(V) complexation) prior the pre-concentration step.

To evaluate the potential matrix effect on the ICP-OES measurement, the matrix-matched calibration was also performed. Three independent four point calibration curves were generated in the range 20–500 μg L⁻¹ by analysis both in HNO₃ (0.5%, v/v) aqueous solution (external calibration) and in the SPE eluates of each beverage. The matrix-effect was evaluated by comparing the respective slopes observed in HNO₃ (0.5%, v/v) aqueous solution and in matrix-matched solution. The slope values resulted not significantly different (p = 0.05), indicating that the matrix effect was negligible. Good linearity was observed, with correlations coefficients (r) higher than 0.9998.

3.4 Application to commercial beverages

The TiO₂/SiO₂-based SPE procedure was applied for determining V(V) in commercial Italian wines (Ortrugo, 11% vol. and Merlot, 12% vol.) and two tea infusions (Earl Grey and Green Blends). V(V) was effectively determined in the commercial wines at concentration levels of some tens of micrograms per liter, namely 26 μg L⁻¹ in Merlot and 80 μg L⁻¹ in Ortrugo (RSD < 5%, n = 3). Differently, V(V) concentration was below MDL in the two tea infusions.

To further assess the reliability of the procedure, four independent aliquots of each sample were spiked with 0, 50, 100 and 250 μg L⁻¹ V(V), pre-concentrated on the ST1 sorbent, and then analyzed. Results obtained by the standard addition method were in excellent accordance with those found by external calibration (differences < 4%), substantiating the quantitative recovery of the metal ion also from these complex matrices. This further gave evidence of the good overall accuracy of the SPE-ICP-OES procedure here presented.