Analytical performance of benchtop total reflection X-ray fluorescence instrumentation for multielemental analysis of wine samples

Spectrochimica Acta Part B 120 (2016) 37–43

Contents lists available at ScienceDirect

Spectrochimica Acta Part B journal homepage: www.elsevier.com/locate/sab

Analytical performance of benchtop total reflection X-ray fluorescence instrumentation for multielemental analysis of wine samples Rogerta Dalipi a, Eva Marguí b,⁎, Laura Borgese a, Fabjola Bilo a, Laura E. Depero a a b

Department of Mechanical and Industrial Engineering, University of Brescia, 25123 Brescia, Italy Department of Chemistry, University of Girona, Campus Montilivi, 17071 Girona, Spain

a r t i c l e

i n f o

Article history: Received 18 September 2015 Received in revised form 1 April 2016 Accepted 6 April 2016 Available online 08 April 2016 Keywords: TXRF Wine Multielemental analysis Mo X-ray tube W X-ray tube

a b s t r a c t Recent technological improvements have led to a widespread adoption of benchtop total reflection X-ray fluorescence systems (TXRF) for analysis of liquid samples. However, benchtop TXRF systems usually present limited sensitivity compared with high-scale instrumentation which can restrict its application in some fields. The aim of the present work was to evaluate and compare the analytical capabilities of two TXRF systems, equipped with low power Mo and W target X-ray tubes, for multielemental analysis of wine samples. Using the Mo-TXRF system, the detection limits for most elements were one order of magnitude lower than those attained using the W-TXRF system. For the detection of high Z elements like Cd and Ag, however, W-TXRF remains a very good option due to the possibility of K-Lines detection. Accuracy and precision of the obtained results have been evaluated analyzing spiked real wine samples and comparing the TXRF results with those obtained by inductively coupled plasma emission spectroscopy (ICP-OES). In general, good agreement was obtained between ICP-OES and TXRF results for the analysis of both red and white wine samples except for light elements (i.e., K) which TXRF concentrations were underestimated. However, a further achievement of analytical quality of TXRF results can be achieved if wine analysis is performed after dilution of the sample with de-ionized water. © 2016 Elsevier B.V. All rights reserved.

1. Introduction The analysis of wine is of great interest, since it is a common alcoholic beverage widely consumed around the world. It has been demonstrated that the daily consumption of wine, contributes significantly to the dietary intake of elements like Ca, Co, Cr, Cu, Fe, K, Mg, Mn, Mo, Ni or Zn which are considered essential for humans [1]. Anyway, an excessive intake of some of the elements mentioned above or of other potentially toxic elements like As, Cd and Pb may be harmful for human health [2]. On this basis, it is obvious that elemental analysis of wine is important for wine-making industry and consumers. In fact, wine constituents are rigorously regulated by International Organization of Vine and Wine [3] and European Commission [4]. Moreover, due to the high complex matrix, the elemental analysis of wine is a challenging task for analytical chemists. Atomic spectrometry techniques such as inductively coupled plasma–mass spectrometry (ICP-MS) [5–8], ICP atomic emission spectrometry (ICP-OES) [9,10], electrothermal atomic absorption spectrometry (ETAAS) [8,11] and flame atomic absorption spectrometry (FAAS) [12] have been widely used for elemental analysis of wine.

⁎ Corresponding author. E-mail address: [email protected] (E. Marguí).

http://dx.doi.org/10.1016/j.sab.2016.04.001 0584-8547/© 2016 Elsevier B.V. All rights reserved.

Another possibility is the use of total reflection X-ray fluorescence spectrometry (TXRF). TXRF is a well established analytical technique for multi element determination in various types of samples, especially liquids and powdered, or micro samples [13]. To perform analysis under total-reflection conditions, samples must be provided as thin films. For liquid samples, this is done by depositing 5–50 μL of sample on a reflective carrier and subsequently drying of the drop. The TXRF system exploits that, at very low glancing angles of the primary X-ray beam (≈ 0.1°), X-ray photons are almost completely absorbed within thin specimens. Therefore, the high background that would generally occur due to scatter from the sample support is absent leading to improved detection limits at μg L−1 level [14]. In the last decades, some papers have been published about wine analysis by TXRF [6,15–22]. These studies are mostly focused on wine contamination from exogenous sources during the wine manufacturing process [15,16,18], wine classification for growing areas [19,20] or production stages [21]. According to Anjos and Castiñeira, matrix effects can be neglected when using TXRF for wine analysis, so only a minimal sample pretreatment is necessary [6,15]. Some other advantages of this technique over ICP-MS are the possibility to get simultaneous multielemental information about the wine sample, the low amount of sample required to perform the analysis and the possibility to get quantitative results without external calibration. Usually, quantitative analysis of wine samples by ICP techniques entails the use of matrix-matched standards for calibration

38

R. Dalipi et al. / Spectrochimica Acta Part B 120 (2016) 37–43

Table 1 Instrumental characteristics and measurement conditions for TXRF and ICP-OES analysis of wine samples. Benchtop TXRF spectrometers (S2 PICOFOX, Bruker AXS Microanalysis)

Instrumental characteristics Anode X-ray tube Maximum power Optics Detector Filter Sample changer Atmosphere Voltage Current Measuring time Instrumental characteristics Element (wavelength) RF Power Plasma gas flow rate Plasma configuration Nebulizer type Wavelength selector Detector

Mo system

W system

Mo Air-cooled metal ceramic 40 W Multilayer monochromator (17.5 keV) Silicon drift detector, Area: 30 mm2, FWHM: 139.43 eV (Mn Kα) Mo 10.00 μm Manual version for single samples Air 50 kV 750 μA 600 s

W Air-cooled metal ceramic 50 W Multilayer monochromator (35 keV) Silicon drift detector, Area: 10 mm2, FWHM: 146.72 eV (Mn Kα) Ni 50.00 μm Automatic version with cassette for up to 25 samples Air 50 kV 1000 μA 2000 s

ICP-OES spectrometer (Agilent ICP-OES 5100-SVDV) K (404.721 nm), Ca (317.933 nm), Mn (257.610 nm), Fe (238.204 nm), Ni (231.604 nm), Cu (327.395 nm), Zn (213.857 nm), Rb (780.026 nm), Sr (407.771 nm), Pb (217.000 nm) 1200 W 12 L min−1 Axial Concentric Polychromator Silicon based multichannel array detector CCD (Charge Coupled Device)

purposes in order to overcome matrix effects [7]. In the case of TXRF, quantification can be performed directly by internal standardization, and thus, the quantification procedure is faster and easier. Moreover, with the recent development and commercialization of benchtop systems, which do not require cooling media or gas consumption for operation, TXRF analysis is also cost-effective compared to other atomic spectroscopic techniques. Nevertheless, benchtop TXRF systems usually have limited sensitivity in comparison with high-scale TXRF instrumentation, which can restrict application in some fields. Most of the contributions published so far dealing with the analysis of wine samples by TXRF analysis were performed using large-scaled systems [15,16, 18–20]. The aim of the present contribution is to evaluate the real analytical capabilities of low power benchtop TXRF instrumentation for multielemental analysis of wine samples. In a recent paper [17], we have explored the capabilities of a benchtop TXRF system, equipped with a low power Mo X-ray tube, for routine multi-element analysis of Italian wines. However, commercial TXRF systems can be equipped with both, Mo and W X-ray tubes and for this reason in the present contribution we have evaluated and compared the possibilities of both TXRF systems, in order to demonstrate the real possibilities of benchtop TXRF systems for the analysis of wine samples.

2. Experimental 2.1. Reagents and solutions Stock solutions of 1000 mg L−1 (ROMIL PrimAg@ Mono-component reference solutions) of interesting elements were used to prepare standard solutions. Ultrapure de-ionized water, used for dilution of stock solutions, samples and preparation of alcoholic synthetic solution (12% ethanol v/v), was obtained from a Milli-Q purifier system (Millipore Corp., Bedford, Massachusetts). ICP multielement standard solution IV (23 elements in 6% nitric acid, Merck KGaA, Darmstadt, Germany) was used to prepare the reference solutions in 12% ethanol v/v matrix. Silicone solution in isopropanol (Serva GmbH & Co, Germany) was used to coat all the quartz glass disc reflectors in order to obtain a hydrophobic film before deposition of liquid samples droplets.

2.2. Samples and synthetic alcoholic standards 2.2.1. Wine samples Nine different Italian wine samples (seven red and two white) were provided by ‘Istituto Zooprofilattico Sperimentale della Lombardia e dell’Emilia Romagna’ of Brescia, Italy. All the wine samples come from a control sampling of commercial wines of the Emilia Romagna Italian Region. 2.2.2. Synthetic alcoholic standards Since no certified reference materials (CRMs) for trace elements determination in wine are available, limits of detection (LOD) for both TXRF systems were evaluated using a synthetic alcoholic solution. Absolute ethanol was added to ultrapure de-ionized water in order to obtain a final alcohol concentration similar to that of a typical wine (12% v/v). The alcoholic solution was added to the ICP multielement standard solution IV to prepare a reference solution with concentrations of 10 mg L−1. Gallium was chosen as internal standard (IS) for quantitative analysis of other elements. 2.2.3. Spiked wine samples The evaluation of the accuracy was performed in a red wine sample by a spiking procedure. The target elements were Ca, Ni, Cd and Pb. Four aliquots were taken from the wine sample solution with a defined volume (about 200 μL). In the first aliquot was not added any standard solution (c0), while the other two aliquots were spiked with a fixed small volume (20 μL) of a standard solution (aqueous solution) containing the analyte elements (Ca, Ni, Cd, and Pb) in different concentrations (c1 and c2). The final solutions were thoroughly mixed to ensure homogeneity. The influence of the sample matrix on the accuracy was also evaluated by spiking the red wine sample after it was diluted in the ratio 1:1 with ultrapure de-ionized water. 2.3. Sample preparation for TXRF analysis The best sample preparation conditions for wine sample analysis by TXRF were carefully evaluated (see Section 3.1). Finally, the selected conditions for wine analysis were as follows: sample solutions were prepared by weighing 1 mL of each wine sample and adding the

R. Dalipi et al. / Spectrochimica Acta Part B 120 (2016) 37–43

appropriate amount of a Ga standard solution at a concentration of 100 mg L−1, to obtain a final Ga concentration of 1 mg L−1. From each wine sample solution containing internal standard, two replicates were prepared by depositing 10 μL on a quartz glass reflector and dried subsequently using an IR lamp. 2.4. TXRF instrumentation and measurement conditions TXRF analysis of the wine samples was performed using two commercial benchtop TXRF systems equipped with Mo and W low-power X-rays tubes (S2 PICOFOX, Bruker AXS Microanalysis GmbH, Berlin, Germany). For comparison purposes, several wine samples were also analyzed by ICP-OES spectrometry (Agilent ICP-OES 5100-SVDV, Agilent Technologies, Spain). To avoid interferences caused by the presence of ethanol in wine samples, standard solutions were prepared in an hydroalcoholic solution (12% (v/v) ethanol) [23]. Spectrometer specifications and final measurement conditions used for TXRF and ICP-OES analysis of wines are summarized in Table 1. 3. Results and discussion Before comparison of the analytical capabilities of Mo and W TXRF systems for wine analysis we explored the best conditions for multielemental analysis of wine samples including sample preparation (internal standardization, sample deposition volume and drying mode) and TXRF measuring conditions (measuring time). If otherwise not indicated, the presented results were obtained using the W-TXRF system. 3.1. Evaluation of wine sample preparation and TXRF measurement 3.1.1. Sample deposition volume To perform analysis under total reflection conditions, samples must be provided as a thin layer (b100 μm) on a carrier with high reflectivity. For liquid samples, usually 5–50 μL are deposited on a carrier and subsequently dried. When performing direct analysis of wine samples, the choice of adequate deposition volume is very important due to the complex matrix of wine. Moreover, the width of the sample spot on the sample carrier has to be within the beam size for complete exposition to the X-ray beam. For this reason, one wine sample was chosen to study the effect of different deposited drop volumes: 5, 5 + 5, 10 and 10 + 10 μL (see Fig. S1 (Appendix A)). It was found that for concentrations usually determined in wine samples, no significant differences regarding the sample volume used for TXRF analysis were observed. As expected, we obtain better sensitivity using higher sample volumes and this is particularly important when we are analyzing trace levels. On this basis, for further experiments, a volume of 10 μL was established as adequate for our working concentrations range. 3.1.2. Internal standardization In TXRF, quantification is usually carried out by internal standardization. This method is based on the addition of an element named internal standard (IS) which is not present in the sample. Commonly, rare elements like Ga or Y are chosen. Medium-heavy elements with K-lines detection are preferred over heavy elements with L-lines detection due to the lower number of peaks. Lighter elements with Z ≤ 21 are not suitable as standards because of particle size-effects that are problematic in low-energy range E b 4 keV [24]. When a W anode is used to generate X-rays, other elements, for example Rh measured by K-lines can be also used as internal standard. We studied Ga (1 mg L−1) as IS for both systems and Rh (5 mg L−1) for W system. No significant differences were observed due to the use of Ga and Rh (see Fig. S2 (Appendix A)). As a consequence, Ga was chosen for further TXRF experiments since it can be used in both Mo and W TXRF systems.

39

3.1.3. Drying mode After depositing the sample solution on a reflector, the drop must be dried before performing TXRF measurements. Usually, liquid samples deposited on quartz reflectors are dried by heating using an IR lamp, hot plate, vacuum or a laminar flow box [25]. We tested all the mentioned drying modes for wine sample solutions. No significant differences were observed on elemental determination with respect to different drying conditions (see Fig. S3 (Appendix A)). However, the time required to dry the drop is significantly reduced when an IR lamp is used (around 2 min) compared to vacuum chamber (more than 1 h) and laminar flow box at room temperature (several hours). Taking into account the obtained results, the IR lamp was considered as the most suitable drying mode for wine samples. 3.1.4. Measuring time In a recent publication we demonstrated that a measuring time of 600 s can be selected for wine analysis when using a low power Mo-TXRF system [17]. However, when using a TXRF system equipped with a W X-ray tube, the measuring time necessary to obtain a similar precision can be different due to the different instrumental sensitivity. In Table 2, relative standard deviation (RSD) values obtained for the analysis by W-TXRF of a wine sample (five replicates) using 600 and 2000 s are displayed. As example, we report RSD values obtained for a major element (K), a minor element (Ca), a trace element (Rb) and an ultratrace element (Mn) present in wine samples. As it is shown, for K and Ca, the impact on precision when longer measuring times are used is not so significant. A different situation is presented for trace and ultra-trace elements in wine like Rb and Mn, where a measuring time of 2000 s leads to an improvement on precision. We studied the influence of measuring time also on LODs (data no reported). A decrease of LODs of about 42% and 30% was obtained when measurements were performed on 2000 s compared to 600 s and 1000 s, respectively. In view of the achieved results, further measurements were performed using a measuring time of 2000 s. 3.2. Evaluation of detection limits Limits of detection were calculated according to [24] by analyzing a multi-element standard solution prepared in water with 12% v/v ethanol (element concentration 10 mg L−1) and a red wine sample (see Table 3 for details on chemical composition of this sample). It was considered appropriate to use this standard in order to evaluate detection limits of other elements and metals that can be present in wine samples but that were not detected in the wine samples studied. Results obtained for LOD of both TXRF systems are shown in Fig. 1. Results showed that limits of detection for light-medium Z elements with K-line detection (K-Sr) and high Z elements with L-line detection (i.e., Ba, Tl-Bi) were one order of magnitude lower when using the Mo-TXRF system (0.01–0.1 mg L−1). On the contrary, better sensitivity was achieved for high Z elements like Ag, Cd and In, when using the W-TXRF system due to the possibility of K-lines detection. These elements are measured by L-line detection with the Mo target, but this is really difficult due to the overlapping peaks of their L-lines with K-lines of K from the sample matrix and Ar that comes from the air. It Table 2 RSD values (%) for five replicates of one red wine sample measured for 600 and 2000 s. Element

Concentration range (mg L−1)

RSD (%) Measurement time (s)

K Ca Rb Mn

600

2000

Real samples

Classification interval

4.5 6.3 10.6 59.0

2.2 4.9 2.4 23.3

254–757 41–101 0.9–3.4 0.5–0.9

Major (100–10,000) Minor (10–100) Trace (1–10) Ultra-trace (b1)

0.002 ±

0.005 0.005 0.05 0.02 0.06 ± ± ± ± ±

0.01 0.02 ± ±

± ±

0.01 0.03 0.02

0.02

± ± ±

±

± ±

0.09 0.07

356a 87a n.d. 0.80 0.35 n.d. 0.019 0.017 1.32 1.04 0.38 n.d. n.d. n.d. n.d. n.d. 0.027 n.d.

Mo system

10 3 ± ± 0.4 0.2

0.001 0.003

0.001 0.001 0.001 0.001 0.001

0.01

± ±

± ± ± ± ±

±

Fig. 1. LOD for both TXRF systems obtained by analyzing a multi-element standard solution prepared in water with 12% v/v ethanol (element concentration 10 mg L−1) and for a red wine sample (see Table 3 for details). For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.

is also interesting to note that limits of detection for light elements in the alcoholic standard are slightly underestimated in comparison with those determined in a real wine sample.

0.001

0.001 0.003 0.02 0.02 0.009

5

0.02 0.06

650.8 87.9 b0.01 0.846 0.461 b0.01 0.029 0.020 1.667 1.068 0.552 b0.01 b0.01 b0.01 b0.01 b0.01 0.04 b0.01

± ±

350 86a n.d. 0.84 0.41 n.d. n.d. n.d. 1.44 1.13 0.45 n.d. n.d. n.d. n.d. n.d. 0.04 n.d.

a

W system

30

White wine sample

ICP-OES (mg L−1)

Measured TXRF value (mg L−1)

4 2

R. Dalipi et al. / Spectrochimica Acta Part B 120 (2016) 37–43

±

± ± ± ± ±

± ±

± ±

320 92a n.d. 0.74 1.14 n.d. 0.020 0.035 1.15 0.95 0.773 n.d. n.d. n.d. n.d. n.d. 0.020 n.d. 0.03 0.001 0.01 ± ± ±

0.02 0.04 ± ±

405 102a n.d. 0.82 0.85 n.d. n.d. n.d. 1.19 1.162 0.89 n.d. n.d. n.d. n.d. n.d. n.d. n.d.

± ±

6 2

0.002 0.001

0.001 0.001 0.001 0.003 0.001

0.005

± ±

± ± ± ± ±

±

0.3 1.2 0.6 0.4 0.1 0.2 0.2 0.6 ± ± ± ± ± ± ± ± 9.9 13.1 6.9 6.0 8.7 8.8 8.5 8.7 0.2 0.1 0.01 0.4 0.3 0.1 0.15 0.2 ± ± ± ± ± ± ± ± 11.2 10.4 10.81 11.1 5.2 7.7 8.13 7.2 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 ± ± ± ± ± ± ± ± 10.01 10.14 10.14 10.02 10.01 10.05 10.05 10.04

Results obtained for the analysis of diluted wine samples (1:1 and 1:3 ratios) are displayed in Fig. 2.

0.1 0.5 ± ±

767.3 120.9 b0.01 1.033 1.022 b0.01 0.026 0.049 1.419 1.079 1.068 b0.01 b0.01 b0.01 b0.01 b0.01 0.023 b0.01 ± ± ± ± ± ± ± ± ± 9 9.0 8.5 8.9 10.0 9.1 9.6 9.4 9.7 2 0.004 1 0.1 0.4 0.1 0.01 0.1 0.02 ± ± ± ± ± ± ± ± ± 15 9.562 11 10.3 10.3 10.3 10.52 10.4 10.28 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 ± ± ± ± ± ± ± ± ± 10.06 9.98 10.02 10.04 10.02 10.01 10.02 10.01 10.02

K Ca Cr Mn Fe Co Ni Cu Zn Rb Sr Ag Cd In Ba Tl Pb Bi

a

Mo system

a a

W system Mo system W system

Expected value (mg L−1)

Measured TXRF value (mg L−1)

1 0.6 0.2 0.2 0.2 0.2 0.2 0.2 0.2

Red wine sample

ICP-OES (mg L−1)

Measured TXRF value (mg L−1)

3.3. Evaluation of accuracy

Alcoholic standard solution

Table 3 Results obtained by using both TXRF systems for a multi-element standard solution prepared in water with 12% v/v ethanol (element concentration 10 mg L−1) and for a red and white samples. For comparison purposes, element concentrations in wine samples determined by ICP-OES are also included. In all cases, results are expressed as mean values of duplicate measurements with the associated standard deviation.

40

In Table 3 are shown the concentration values obtained for TXRF analysis of the synthetic alcoholic solution aforementioned and also for two wine samples. For comparison purposes, element concentrations in wine samples determined by ICP-OES are also included. As it was expected, taking into account the results obtained in Fig. 1, for the synthetic alcoholic solution the W system provides better recoveries for Ag, Cd and In, while with the Mo system we have better recoveries for lights Z elements (such as K) and the heaviest ones as Ba, Tl, Pb and Bi. This is due to the fact that these latter elements are measured by their L-lines in both systems, but the Mo system provides better sensitivities and so the LODs are lower. Therefore, a complementary use of both TXRF systems may be very useful if we are interesting on accurate results of target elements in wine (i.e. Cd and Pb). As it is shown in Table 3, in general, good agreement was obtained between ICP-OES and TXRF results for the analysis of red and white samples. The only exception was the case of elements with a low atomic number (K and Ca) which concentrations are underestimated by TXRF analysis probably due to absorption issues from wine matrix. In order to investigate in more detail wine matrix effects, a spiking procedure with a diluted and a non-diluted red wine (see Section 2.2.3 for details) was conducted. Wine sample solutions were spiked with a fixed small volume of a standard solution containing analyte elements (Ca, Ni, Cd, and Pb) in different concentrations (c1 and c2) and measured by both TXRF systems. We decided to perform this procedure for one light Z element (Ca), one medium-heavy Z element (Ni), one heavy Z element (Cd) with K-line detection in W system and L-line detection in Mo system and one heavy Z element (Pb) with L-line detection in both systems. In Table 4 the obtained concentrations and recoveries for the mentioned elements in the considered diluted and not diluted wine solutions are shown. As it can be seen, for the case of Ca the recoveries obtained in all the three additions are improved significantly (7–27%) in the diluted wine by measuring with both TXRF systems. In the case of Ni by measuring with W system, good recoveries are obtained for the 1st and the 2nd addition and there are no significant differences between not diluted and diluted wine samples. On the other way with the Mo system, the recoveries are improved (12–16%) in the diluted wine. Cadmium was not present in wine samples, but the aim was to compare the K-line and L-line detection by employing the W and Mo target X-ray tube respectively. The determination of Cd with the Mo system is limited because of low sensitivity and interferences of

R. Dalipi et al. / Spectrochimica Acta Part B 120 (2016) 37–43

41

Table 4 Calculated and measured concentrations (mg L−1) and the respective recoveries (%) for Ca, Ni, Cd and Pb in the spiked diluted and not diluted red wine sample. Analyte

Ca

Ni

Cd

Pb

Type of sample

Not diluted Diluted Not diluted Diluted Not diluted Diluted Not diluted Diluted Not diluted Diluted Not diluted Diluted Not diluted Diluted Not diluted Diluted Not diluted Diluted Not diluted Diluted Not diluted Diluted Not diluted Diluted

Addition

C0 (no) C1 (1st) C2 (2nd) C0 (no) C1 (1st) C2 (2nd) C0 (no) C1 (1st) C2 (2nd) C0 (no) C1 (1st) C2 (2nd)

Addition concentration (mg L−1)

0 0 57 56 84 90 0 0 0.54 0.53 0.83 0.85 0 0 0.53 0.53 0.82 0.84 0 0 0.54 0.53 0.83 0.85

W system

Mo system

Found concentration (mg L−1)

Recovery (%)

79 54 131 109 149 135 n.d. n.d. 0.53 0.53 0.76 0.84 n.d. n.d. 0.45 0.48 0.74 0.84 n.d. n.d. 0.45 0.47 0.62 0.70

– – 92 100 84 91 – – 98 100 92 90 – – 85 91 90 100 – – 83 88 74 82

Found concentration (mg L−1) 75 59 115 104 137 143 0.02 0.01 0.48 0.55 0.73 0.84 n.d. n.d. n.d. n.d. n.d. n.d. 0.02 0.01 0.35 0.41 0.56 0.66

Recovery (%) – – 71 81 75 93 – – 84 100 85 97 – – – – – – – – 63 75 66 76

Recovery values were calculated as follows: Recovery ð%Þ ¼

ðFound concentration ðC1 or C2Þ  Found concentration ðC0ÞÞ  100: Addition concentration

Cd L-line with K-line of K that is the major element in wine samples. In fact, Cd could not be detected in none of the wine samples measured by the Mo system, while good recoveries (85–104%) were obtained with the W system.

Lead is measured by using its L-lines in both TXRF systems. Recoveries ranged from 74 to 112% by employing the W system and from 63 to 76% by the Mo system. Even for this element, better recoveries are obtained for the diluted wine samples, demonstrating the influence of

Fig. 2. Effect of wine sample dilution on element concentrations. For more information about red and wine samples composition see Table 3. Legend: W (TXRF with W X-ray tube), Mo (TXRF with Mo X-ray tube).

42

R. Dalipi et al. / Spectrochimica Acta Part B 120 (2016) 37–43

the sample matrix on the determination of elements which are measured through L-lines. Taking into account these results it was considered appropriate to study in more detail the effect of wine sample dilution on element concentrations. For that, a red and a white wine sample (see Table 3 for details) were diluted with deionized water using 1:1 and 1:3 ratios. In Fig. 2, concentration results obtained for the analysis of diluted samples as well as undiluted samples by TXRF and ICP-OES are displayed for two low-Z elements (K and Ca) and two trace elements (Mn and Zn). As it is shown, it is evident that for the determination of low-Z elements (above all for K) it is necessary to dilute wine samples using a ratio higher than 1:1. In this case, this is not a problem since K and Ca are present at high concentration levels (hundreds of mg L−1) in wine samples. On the contrary, for minor rand trace elements such as Mn and Zn a good strategy is to measure the wine samples without dilution or with a dilution 1:1 (depending on the concentration level).

Fig. 3. TXRF raw spectra for the analysis of a red wine sample obtained using Mo and W excitation source. For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.

3.4. Evaluation of precision One red wine sample was chosen to evaluate the precision at different levels such as global precision, drop deposition uncertainty and instrumentation uncertainty. For the global precision, four sample solutions at the same conditions were prepared. In about 200 μL of wine, we added 20 μL of 10 mg L−1 Ga standard solution, to have a final Ga concentration about 1 mg L−1. From each solution, 10 μL were deposited on four separate quartz glass reflectors. The uncertainty of drop deposition was evaluated by depositing 10 μL of one sample solution prepared previously for the global precision in four different reflectors. In this case, the instrument uncertainty is also included. For the evaluation of instrument related uncertainty, one reflector containing 10 μL of sample solution was measured four times in the same conditions, but with different positions of the quartz reflector, to include the influence of deposition in-homogeneity. In all cases, RSD of the obtained results were calculated. It was found that for lighter elements (K and Ca) the major contribution on the global uncertainty is mostly due to sample preparation. Otherwise for the other detected elements (Mn, Fe, Zn, Br, Rb and Sr) instrument uncertainty had a major role in the global uncertainty of the obtained results. It is also interesting to note that the global precision is slightly lower for the Mo-TXRF system (RSD ~ 3%) in comparison with the W-TXRF system (RSD ~ 11%). These values are acceptable if we take into account the difficult matrix of wine samples. 3.5. Analysis of red and white wines In order to evaluate if there is any correlation between the W and Mo TXRF data we performed a linear regression analysis with all the results obtained using both TXRF systems. In Table 5 are shown the concentration range of elements (mg L−1) and values of parameters for the linear regression. As we can see, a good agreement was obtained for

Table 5 Concentration range of elements (mg L−1) in all the studied wine samples and linear regression (LR) parameters between W and Mo TXRF datasets. Element

K Ca Mn Fe Cu Zn Br Rb Sr

Concentration range (mg L−1)

№ of samples

Parameters for LR R2

A

=0

B

=1

254–757 41–101 0.5–0.9 0.3–3.8 0.02–0.37 0.1–1.4 0.05–1.19 0.9–3.4 0.3–1.2

9 9 9 9 4 9 9 9 9

– – – 0.973 0.996 0.991 0.909 0.974 0.913

– – – 0.096 0.007 0.057 0.016 −0.034 −0.025

– – – Yes Yes Yes Yes Yes Yes

– – – 0.8713 1.1464 1.0290 0.9439 1.2285 1.3075

– – – Yes Yes Yes Yes Yes Yes

Regression analysis: LR model, TXRF (W) = A + B TXRF (Mo) (A: Intercept, B: Slope).

concentration values of elements from Fe to Sr determined by W TXRF system with those determined by Mo TXRF system and no statistical differences at 95% confidence level were found. For K, Ca and Mn the results were not satisfactory as any linear correlation between W and Mo TXRF data could not be obtained. As above explained, this is due to sensitivity issues and higher detection limits when W target X-ray tube is employed. As an example, in Fig. 3, typical raw spectra of a red wine sample measured by both TXRF systems are displayed. As it is shown, the main differences between Mo and W excitation are regarding the measurement energy (keV) range and the produced background. In both TXRF spectra Si K-lines from quartz glass sample carrier and Ar K-lines from the atmospheric air (measurements are not performed in vacuum conditions) are detected. The Kα and Kβ peaks of IS (Ga) are also observed. In Mo TXRF spectrum, the strong peak around 17 keV is due to the elastic (Mo Kα) and inelastic (Compton) scattering of the primary X-ray source used for excitation of the sample. On the other way, a bremsstrahlung continuum radiation is produced when a W target is used to generate X-rays.

4. Conclusions In this paper we evaluate the possibilities and drawbacks of lowpower benchtop TXRF instrumentation for element determination in wine samples. For routine and screening analysis of red and white wines, sample analysis can be directly performed depositing 10 μL of the internal standardized sample (using Ga at a concentration of 1 mg L− 1) on a quartz glass reflector and using a measuring time of 600 s (Mo-TXRF systems) or 2000 s (W-TXRF systems). Using such approach, good agreement was obtained between ICP-OES and TXRF results for the analysis of both red and wine samples. Only light element concentrations (i.e., K and Ca) were underestimated when performing wine analysis by TXRF. However, the accuracy study performed by a spiking procedure of real diluted and not diluted wine samples demonstrated that a further achievement of analytical quality of TXRF results can be achieved if wine analysis is performed after dilution of the sample with de-ionized water (1:1). Additional advantages of the TXRF method proposed are the simultaneous multielemental information of the sample, the easy quantification through internal standardization and also low operating costs, since the benchtop system use does not require cooling media and gas consumption for function. Taking into account the specific analytical capabilities of Mo-TXRF (better for light-medium Z elements with K-line detection and high Z elements with L-line detection) and W-TXRF systems (better for high Z elements with K-line detection) we propose a complementary use of both TXRF configurations to detect the broad range of elements present in wines, including some potentially toxic elements such as Cd and Pb.

R. Dalipi et al. / Spectrochimica Acta Part B 120 (2016) 37–43

Acknowledgments This work was supported by the Spanish National Research Program (Project ref. CGL2013-48802-C3-2-R). The authors would like to acknowledge Dr. Emanuele Sangiorgi of Istituto Zooprofilattico Sperimentale della Lombardia ed Emilia Romagna, Brescia, Italy. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.sab.2016.04.001. References [1] G. Grindlay, J. Mora, L. Gras, M.T.C. de Loos-Vollebregt, Atomic spectrometry methods for wine analysis: a critical evaluation and discussion of recent applications, Anal. Chim. Acta 691 (2011) 18–32, http://dx.doi.org/10.1016/j.aca.2011.02. 050. [2] B. Tariba, Metals in wine — impact on wine quality and health outcomes, Biol. Trace Elem. Res. 144 (2011) 143–156, http://dx.doi.org/10.1007/s12011-011-9052-7. [3] OIV, Compendium of International Methods of Wine and Must Analysis, Paris, 2015. [4] EEC, COMMISSION REGULATION (EEC) No 000/90 of 17 September 1990 determining Community methods for the analysis of wines, 1990. [5] C.M.R. Almeida, M.T.S.D. Vasconcelos, M. Barbaste, B. Medina, ICP-MS multi-element analysis of wine samples — a comparative study of the methodologies used in two laboratories, Anal. Bioanal. Chem. 374 (2002) 314–322, http://dx.doi.org/10.1007/ s00216-002-1467-8. [6] M.M. Castiñeira, R. Brandt, a von Bohlen, N. Jakubowski, Development of a procedure for the multi-element determination of trace elements in wine by ICP-MS, Fresenius J. Anal. Chem. 370 (2001) 553–558, http://dx.doi.org/10.1007/ s002160100862. [7] V.S. Šelih, M. Šala, V. Drgan, Multi-element analysis of wines by ICP-MS and ICP-OES and their classification according to geographical origin in Slovenia, Food Chem. 153 (2014) 414–423, http://dx.doi.org/10.1016/j.foodchem.2013.12.081. [8] G. Grindlay, J. Mora, L. Gras, M.T.C. de Loos-Vollebregt, Ultratrace determination of Pb, Se and As in wine samples by electrothermal vaporization inductively coupled plasma mass spectrometry, Anal. Chim. Acta 652 (2009) 154–160, http://dx.doi. org/10.1016/j.aca.2009.05.020. [9] A.E. Martin, R.J. Watling, G.S. Lee, The multi-element determination and regional discrimination of Australian wines, Food Chem. 133 (2012) 1081–1089, http://dx. doi.org/10.1016/j.foodchem.2012.02.013. [10] I. Sen, F. Tokatli, Characterization and classification of Turkish wines based on elemental composition, Am. J. Enol. Vitic. 65 (2013) 134–142, http://dx.doi.org/10. 5344/ajev.2013.13081.

43

[11] İ.M. Alkış, S. Öz, A. Atakol, N. Yılmaz, R.E. Anlı, O. Atakol, Investigation of heavy metal concentrations in some Turkish wines, J. Food Compos. Anal. 33 (2014) 105–110, http://dx.doi.org/10.1016/j.jfca.2013.11.006. [12] D. Schiavo, J.Y. Neira, J.A. Nóbrega, Direct determination of Cd, Cu and Pb in wines and grape juices by thermospray flame furnace atomic absorption spectrometry, Talanta 76 (2008) 1113–1118, http://dx.doi.org/10.1016/j.talanta.2008.05.010. [13] E. Marguí, R. Van Grieken, X-ray Fluorescence Spectrometry and Related Techniques: An introduction, Momentum Press, New York, 2013, http://www.rsc.org/ chemistryworld/2013/11/x-ray-fluorescence-spectrometry-xrf-margui-grieken. [14] C. Streli, Recent advances in TXRF, Appl. Spectrosc. Rev. 41 (2006) 473–489. [15] M.J. Anjos, R.T. Lopes, E.F.O. De Jesus, S. Moreira, R.C. Barroso, C.R.F. Castro, Trace elements determination in red and white wines using total-reflection X-ray fluorescence, Spectrochim. Acta B At. Spectrosc. 58 (2003) 2227–2232, http://dx.doi.org/ 10.1016/j.sab.2003.07.004. [16] M.L. Carvalho, Study of Heavy Metals in Madeira Wine by Total Reflection X-ray Fluorescence Analysis, Vol. 251996 29–32. [17] R. Dalipi, L. Borgese, A. Zacco, K. Tsuji, E. Sangiorgi, R. Piro, et al., Determination of trace elements in Italian wines by means of total reflection X-ray fluorescence spectroscopy, Int. J. Environ. Anal. Chem. 1–11 (2015), http://dx.doi.org/10.1080/ 03067319.2015.1036861. [18] S. Galani-Nikolakaki, N. Kallithrakas-Kontos, A.A. Katsanos, Trace element analysis of Cretan wines and wine products, Sci. Total Environ. 285 (2002) 155–163, http://dx. doi.org/10.1016/S0048-9697(01)00912-3. [19] X. Gruber, P. Kregsamer, P. Wobrauschek, C. Streli, Total-reflection X-ray fluorescence analysis of Austrian wine, Spectrochim. Acta B At. Spectrosc. 61 (2006) 1214–1218, http://dx.doi.org/10.1016/j.sab.2006.08.006. [20] S.J. Haswell, A.D. Walmsley, Multivariate data visualisation methods based on multielemental analysis of wines and coffees using total reflection X-ray fluorescence analysis, J. Anal. At. Spectrom. 13 (1998) 131–134. [21] S. Pessanha, M.L. Carvalho, M. Becker, A. Von Bohlen, Quantitative determination on heavy metals in different stages of wine production by total reflection x-ray fluorescence and energy dispersive x-ray fluorescence: comparison on two vineyards, Spectrochim. Acta B At. Spectrosc. 65 (2010) 504–507, http://dx.doi.org/10.1016/j. sab.2010.04.003. [22] S. Kunimura, J. Kawai, Application of a portable total reflection X-ray fluorescence spectrometer to a trace elemental analysis of wines, Bunseki, Kagaku 58 (2009) 1041–1045. [23] T. Corona, M. Iglesias, E. Anticó, Migration of components from cork stoppers to food: challenges in determining inorganic elements in food simulants, J. Agric. Food Chem. 62 (2014) 5690–5698. [24] R. Klockenkämper, A. von Bohlen, Total-reflection X-ray Fluorescence Analysis and Related Methods, 2nd edition, 2015. [25] I. De La Calle, N. Cabaleiro, V. Romero, I. Lavilla, C. Bendicho, Sample pretreatment strategies for total reflection X-ray fluorescence analysis: a tutorial review, Spectrochim. Acta B At. Spectrosc. 90 (2013) 23–54, http://dx.doi.org/10.1016/j. sab.2013.10.001.