Atomic spectrometry methods for wine analysis: A critical evaluation and discussion of recent applications

Analytica Chimica Acta 691 (2011) 18–32

Contents lists available at ScienceDirect

Analytica Chimica Acta journal homepage: www.elsevier.com/locate/aca

Review

Atomic spectrometry methods for wine analysis: A critical evaluation and discussion of recent applications Guillermo Grindlay a,∗ , Juan Mora a , Luis Gras a , Margaretha T.C. de Loos-Vollebregt b a b

Department of Analytical Chemistry, Nutrition and Food Sciences, University of Alicante, PO Box 99, 03080 Alicante, Spain Delft University of Technology, Faculty of Applied Sciences, Analytical Biotechnology, Julianalaan 67, 2628 BC Delft, The Netherlands

a r t i c l e

i n f o

Article history: Received 12 January 2011 Received in revised form 15 February 2011 Accepted 17 February 2011 Available online 24 February 2011 Keywords: Atomic spectrometry Wine Elemental analysis Matrix effects

a b s t r a c t The analysis of wine is of great importance since wine components strongly determine its stability, organoleptic or nutrition characteristics. In addition, wine analysis is also important to prevent fraud and to assess toxicological issues. Among the different analytical techniques described in the literature, atomic spectrometry has been traditionally employed for elemental wine analysis due to its simplicity and good analytical figures of merit. The scope of this review is to summarize the main advantages and drawbacks of various atomic spectrometry techniques for elemental wine analysis. Special attention is paid to interferences (i.e. matrix effects) affecting the analysis as well as the strategies available to mitigate them. Finally, latest studies about wine speciation are briefly discussed. © 2011 Elsevier B.V. All rights reserved.

Guillermo Grindlay was born in Granada (Spain) in 1978. He studied chemistry at the University of Alicante where he obtained his PhD in 2006 working on a microwave fully based sample introduction system atomic spectrometry. He is currently an assistant professor in the University of Alicante. Over the past few years, he has performed several post-doctoral research works at TUDelft and the University of Ghent. His current research interests are focused on the development of new sample introduction systems for plasma based techniques as well as on the origin of non-spectral matrix effects.

∗ Corresponding author. Tel.: +34 965903400; fax: +34 965903527. E-mail address: [email protected] (G. Grindlay). 0003-2670/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2011.02.050

Juan Mora obtained his PhD degree in Analytical Chemistry from the University of Alicante (Spain) in 1994. He continued his research work as a post-doctoral fellow at the Delft University of Technology (The Netherlands). Currently, he is teaching Analytical Chemistry and member of the Analytical Atomic Spectrometry Group at the Department of Analytical Chemistry, Nutrition and Food Sciences of the University of Alicante. His research interests are in the field of analytical atomic spectrometry and include the development and characterization of different sample introduction systems and the study of fundamental and applied aspects in plasma based atomic spectrometry techniques.

G. Grindlay et al. / Analytica Chimica Acta 691 (2011) 18–32

Luis Gras was born in Elche (Spain) in 1968 and obtained his PhD degree in Chemistry in 1997, after which, he performed a post-doctoral stage at the TUDelft University (Holland). Since 1993 he is part of the Analytical Atomic Spectrometry Group of the University of Alicante and nowadays his main research areas are the design of new instrumentation based on microwave radiation for on-line sample pretreatment and the development of analytical methods for the elemental analysis of foods and the use of chemometrics for sample characterization.

1. Introduction Wine is an alcoholic beverage of a great social and economic significance. This beverage was initially used as mind-altering substance for religious ceremonies in ancient times but, over the years, it has become a standard item in human diet. Nowadays, worldwide wine production and consumption has been estimated over 20,000 million litres per year by the Organisation Internationale de la Vigne et du Vin (OIV) which points out the importance of this product [1]. From a chemical point of view, wine is a complex water–ethanol mixture which contains a great variety of both organic and inorganic substances [2,3]. A brief overview on the chemical composition of wine is shown in Table 1. Organic substances in wine can be divided in two groups: volatile and non-volatile compounds. Among the volatile group, ethanol is the most abundant compound ranging from 8% to 19% (v/v). There are other volatile compounds (e.g. methanol, esters and terpenes) which are directly responsible for wine organoleptic properties but their concentration levels are rather low compared to ethanol. Non-volatile organic compounds include low-volatile alcohols, sugars and organic acids as well as their conjugated salts. They may be present at concentrations above 1.0 g L−1 . Wine also contains small quantities of other substances (<1.0 g L−1 ) such as amino acids, polyphenols, flavonoids, etc. As regards the inorganic fraction, wine is rich in Cl− , PO4 3− , SO4 2− and SO3 2− salts. The most abundant counter ions (i.e. major elements) for both inorganic and organic salts are those related to grape physiological processes such K, Ca, Na and Mg. Among them, K shows the highest concentration levels between 500 and 1500 mg L−1 followed by Ca, Mg and Na around 10–200 mg L−1 . The elements usually present in concentrations ranging from 0.1 to 10 mg L−1 (i.e. trace elements) are Al, B, Cu, Fe, Mn, Rb, Sr and Zn. Finally, ultratrace elements are those below 0.1 mg L−1 such as Se, Pb and Cd. The great number of parameters affecting wine quality has initiated the development of different protocols for analysis [4]. In fact, wine constituents are strictly regulated by international organizations [5] or government agencies [6] to avoid fraud and health risks. Luque de Castro et al. [7] have recently reviewed methods of analysis for the most commonly determined parameters in wine such as ethanol, sulphur dioxide, reducing sugars, polyphenols, organic acids, total and volatile acidity, Fe, soluble solids, pH and color. Wine elemental composition provides additional important information about the quality or characteristics of the wine [8,9]. Pohl has summarized what metals tell us about wine in an article focused on the role of metals in wine and processes involved in winemaking [10]. Elements in wine can be classified into two groups: endogenous and exogenous. Wine endogenous elements, the most abundant, are related with the type of soil in the vineyard, the grape variety and maturity and the climatic conditions. Exogenous elements are associated with the external impurities

19

Margaretha de Loos-Vollebregt graduated from Delft University of Technology and obtained her PhD from the same university in 1980 (supervisor Prof. Leo de Galan). She is currently teaching spectroscopy in the Analytical Biotechnology group at Delft University of Technology and she is guest professor in the Atomic and Mass Spectrometry group of the Department of Analytical Chemistry at Ghent University, Belgium. She is editor of Spectrochimica Acta Part B: Atomic Spectroscopy, advisory board member of Journal of Analytical Atomic Spectrometry and editorial board member of Analytica Chimica Acta. Her research interests are in the field of analytical atomic spectroscopy.

which may reach wine during the growth of grapes. Thus, differences in the content of K, Ca or Cu in different wines can be due to the use of fertilizers for cultivation. Application of pesticides, fungicides and fertilizers during the growing seasons of wines can lead to an increased amount of Cd, Cu, Mn, Pb or Zn in the resulting wine. The environmental pollution of the vineyards may also enhance the concentration of some elements [11]. Thus, wines from vineyards close to the coast show a relatively high concentration of Na whereas high concentrations of Pb or Cd are found in wines from vineyards located close to road traffic or industrial areas. In addition, exogenous metals are related with winemaking (from harvesting to bottling and cellaring) such as: (i) process equipment. The long contact of wine with materials such as Al, brass, glass, stainless steel, wood, etc., used to build pipes, casks and barrels is the usual source for Al, Cd, Cr, Cu, Fe and Zn; or (ii) the addition of different substances at different steps of wine production [12]. Thus, contamination with Na, Ca or Al can be associated with the use of fining and clarifying substances (flocculants such as bentonites) added to the wine to remove suspended solids after fermentation and to reduce turbidity. Ca contamination can also arise from the addition of CaCO3 or CaSO4 or de-acidification of must and wine or enhancement of acidity of grape juices, respectively. The main source of Cu in wine is the CuSO4 added to remove H2 S and other sulphidic compounds. From the statements made above, it is obvious that monitoring elemental wine composition has a great importance for winemaking industry as well as customers. Fig. 1 summarizes the main objectives accomplished with wine elemental analysis. Wine elemental analysis has been employed in the literature for different purposes:

1. Bioavailability/toxicity. It has been demonstrated that daily consumption of wines in moderate quantities significantly contributes to the requirements of the human organism for essential elements such as Ca, Co, Cr, Cu, Fe, K, Mg, Mn, Mo, Ni or Zn, among others [13,14]. In fact, it can be assumed that wine constitutes a very significant contribution to the total diary intake of elements such as V [15] or Al [16]. Its nutritional effect depends on the physicochemical form and concentration of the elements. In fact, some of the above mentioned essential elements are harmful for human health in case of excessive intake. To prevent health damage, wine analysis on the metals’ content and tolerable concentrations are regulated. In addition, according to the different health-protection laws in different countries, it is required to detect the presence of hazardous species (e.g., Cd, Pb, Hg, Al, Tl, As, Sb, S, different organometallic compounds of Pb, As, etc.) [17,18]. Hence, there is an increasing interest in wine speciation analysis to evaluate the toxicity, bioavailability, bioaccumulation and transport of specific elements (Pb, As).

20

G. Grindlay et al. / Analytica Chimica Acta 691 (2011) 18–32

Table 1 Main components found in wine. Wine composition

Concentration

Volatile organics Non-volatile organics

Inorganic salts Elements

Non-volatile alcohols Sugars Organic acids/salts Other substances Major Trace Ultratrace

Ethanol

8–19% (v/v)

Glycerol, butylethylglycol Glucose, fructose, galactose, mannose Tartaric, malic, citric, acetic Aminoacids, polyphenols, flavonoids, etc. Cl− , PO4 3− , SO4 2− , SO3 2− Na, Mg, K, Ca B, Al, Mn, Fe, Cu, Zn, Sr, Rb Li, Sc, Ti, V, Cr, Co, Ni, As, Se, Mo, Ag, Cd, Sn, Sb, Ba, rare earth, Hg, Tl, Pb, etc.

1–10 g L−1 1–200 g L−1 1–8 g L−1 <1 g L−1 >10 mg L−1 >10 mg L−1 0.1–0.1 mg L−1 <0.1 mg L−1

2. Quality assurance/control. Metal analysis is an important tool in the quality control of wines and for the evaluation of the winemaking process [19]. The elemental content may change during wine production [20] and therefore elemental analysis at different points in the winemaking process can provide very useful information. In addition, elements can influence the flavor, taste, color and long-term stability of the final product. Thus, some elements are important for efficient alcoholic fermentation. Ca, K, Mg and Na regulate the cellular metabolism of yeast maintaining the adequate pH and ionic balance. Cu, Fe, Zn and Mn are responsible for changes in stability, color and clarity of old wine and modification of the sensory quality of wine after bottling [10]. 3. Origin, provenance and authenticity. Elemental analysis makes it possible to determine wine authenticity and to detect frauds or adulterations by blending wines by means of its elemental composition. 4. Finally, it is interesting to note that elemental analysis of wines is a challenge for the analytical chemist mainly due to the high matrix complexity. For this reason, wine is often employed as a test sample in the research laboratories when evaluating a new spectroscopic method or part of it. Thus, wine has been used to evaluate several sample introduction devices for spectrometric techniques, sample pretreatment methods, interferences studies, calibration strategies, etc. Various analytical methods have been used for the determination of trace metals in wine such as, e.g., near infrared spectrometry

[21], electrophoresis chromatography [22], ionic chromatography ´ [23,24] or cathodic stripping voltametry [23]. Pyrzynska [25] in 2004 and Aceto et al. [26] in 2002 have reviewed the different methods used for the determination of selected metals in different types of wine and their concentration ranges. Furthermore, Yip et al. [27] discussed commonly used methodologies for inorganic analysis of ˜ wine and Ibanez et al. [9] presented a broad overview of metal determination in alcoholic beverages, including wine analysis. In the present review, we will discuss in detail atomic spectrometry techniques reported for elemental wine analysis and provide a critical evaluation of recent applications in this field. In addition, we will address typical problems related to the complex wine matrix and indicate possibilities how to deal with them. 2. Atomic spectrometry techniques for elemental wine analysis: possibilities and limitations The determination of trace and ultratrace elements in wines by means of atomic spectrometry techniques is routinely carried out in many analytical laboratories but research work in this field is still published every year. In fact, almost 10% of elemental wine analysis literature deals with these techniques. Fig. 2 shows in a diagram which atomic spectrometry techniques have been mostly employed for elemental wine analysis in the literature from 2000 onwards. Among them, 40% of all references are focused on inductively coupled plasma mass spectrometry (ICPMS) [28–33] followed by inductively coupled plasma atomic emission spectrometry (ICPAES) [34–38], electrothermal atomic absorption

Fig. 1. Overview of topics and elements of interest in elemental analysis of wine.

G. Grindlay et al. / Analytica Chimica Acta 691 (2011) 18–32

Fig. 2. Atomic spectrometry techniques employed for elemental wine analysis (source: Scifinder® ; Keywords: wine analysis, english, journal, review, 1999, element, atomic absorption, flame, electrothermal, hydride generation, plasma, emission, mass, fluorescence, X-ray).

spectrometry (ETAAS) [39–43] and flame atomic absorption spectrometry (FAAS) [19,44,45] with 18%, 16% and 15%, respectively. Then, less popular techniques (i.e. <7%) include total reflection X-ray fluorescence (TXRF) spectrometry [46–48], and volatile compound generation-based techniques, i.e. hydride generation atomic absorption spectrometry (HGAAS) [49–52], hydride generation atomic fluorescence spectrometry (HGAFS) [52,53] and cold vapor atomic absorption spectrometry (CVAAS) [49]. Different criteria should be defined to select the appropriate technique for a given application. In wine analysis, main parameters that should be taken into account are: (i) analyte concentration; (ii) number of elements to be determined; and (iii) interferences. 2.1. Analyte concentration As it has been pointed out previously, elemental composition in wine ranges from major to ultratrace levels. For this reason, detection techniques should be selected accordingly to the concentration level of the analytes of interest. The determination of major elements and some trace elements is easily accomplished by FAAS. This technique has been used as the standard in most OIV [5] or European Union [6] methods for elemental wine analysis. Sample dilution allows to bring the concentration of Ca and Mg in the range normally covered by FAAS (0.1–10 mg L−1 ) [19] whereas Fe, Cu and Mn [19,45] can be measured directly. Na and K determination is not commonly performed by FAAS since flame atomic emission spectrometry provides better analytical figures of merit [1,4,6]. Total sulphur content in wines has also been recently determined by continuum source FAAS, using the molecular absorption band of carbon monosulfide [44]. To this end, a high-resolution continuum source atomic absorption spectrometer equipped with a novel high intensity short-arc xenon lamp and an ordinary air-acetylene flame was used. Despite the relatively high limits of detection of FAAS, the analysis of ultratrace elements is still feasible after a preconcentration step [54–59]. Thus, for instance, Tuzen et al. [55,56] measured heavy metals in wine (i.e. Cd, Co, Cr, Ni, Pb, etc.) after a preconcentration procedure with solid phase extraction (SPE) with Diaion HP-2MG or Chromosorb 108 adsorbents. Similar approaches were followed by Lemos et al. [58] and Baircioglu et al. [59] but they coupled a SPE minicolumn to a flow injection (FI) device for on-line analysis. On the other hand, FAAS detection capabilities can also be improved by using an alternative sample introduction system rather than the conventional configuration (i.e. concentric pneumatic nebulizer attached to a spray chamber). Schiavo et al. [60]

21

employed thermospray flame furnace atomic absorption spectrometry for the determination of Cd, Cu and Pb in wine and grape juices. TXRF has also been employed for the analysis of major and some trace elements because it has some attractive features when compared to FAAS such as the low amount of sample required and minimum sample preparation [46]. In addition, this technique offers possibilities for P, Cl and S determination, whereas these elements are difficult to quantify by other atomic spectrometry techniques [48]. However, detection of light elements (atomic mass < 23 amu) is not easy and special instrumentation is required [32]. When simultaneous multielemental quantification of elements in a broad range of concentrations is required, ICPAES is a more appropriate technique [35–37,61]. The possibility to use both axial and radial viewing modes provides a dynamic range from 0.01 to 1000 mg L−1 , which also allows the determination of elements at ultratrace levels. Thus, for instance, Li, Fe, Mn, Cu, Zn, Se, Ba, Pb, Bi, V and even rare earth elements were successfully determined by ICPAES [37,38,62,63]. Similar to FAAS, ICPAES analytical figures of merit can also be further improved by modifying the sample introduction system [13,35] Grindlay et al. [35] have employed a microwave based desolvation system for elemental wine determination which improves limits of detection (LOD) when compared to a conventional sample introduction system. Similar results were also obtained using an ultrasonic nebulizer [13]. The analysis of ultratrace elements is usually performed by means of ETAAS, HGAAS/HGAFS and ICPMS. ETAAS has some attractive features since the wine matrix can be partly removed during the pyrolysis step with the aid of chemical modifiers. Among the elements of interest determined by ETAAS are Pb [39,40,42,43,53,64], Cd [39,40,43,64], As [43,65] Sb [65], Al [13], Cu [43], Cr [39], V [15] and Hg [41]. HGAAS/HGAFS techniques are employed for elements such as Pb [50,51,53,66], As [52,66,67], Bi [66], Sn [66], Se [49] and Te [66]. These techniques offer better sensitivity in comparison with FAAS and faster analysis in comparison with ETAAS. However, literature devoted to ultratrace analysis in wine is mainly focused on ICPMS due to its high sensitivity and broad dynamic range [28–30,32,33,69]. In fact, current ICPMS applications in this field cover almost the whole periodic table from light elements [69,70] to rare earth [71,72] or heavy elements (i.e. Pb or U) [20,28,68]. It is interesting to point out that when precision and/or accuracy are not the critical parameters, ICPMS instruments are used in the semi-quantitative mode with errors lower than 20% [29,68]. 2.2. Mono/multielement capabilities Before a given technique is selected for elemental wine analysis, it is important to check whether the analytes of interest can be measured simultaneously or only sequentially in order to optimize the analysis throughput. Atomic absorption or atomic fluorescence based techniques (i.e. FAAS, ETAAS and HGAAS/HGAFS) are sequentially in nature and, as a consequence, sample throughput is low. On the other hand, TXRF, ICPAES and ICPMS afford multielement analysis capabilities. Thus, for instance, up to 13 elements (P, S, Cl, Ca, Ti, Cr, Mn, Fe, Ni, Cu, Zn, Rb, and Sr) in Brazilian wines were simultaneously quantified by Anjos et al. [48] using TXRF. Similar results were also reported by ˜ Castineira et al. [32] but in their study K was also included among the analytes of interest. As regards plasma based techniques, 38 elements were simultaneously determined by Gonzálvez et al. [38] ˜ in Spanish wines by means of ICPAES whereas Castineira et al. [20] determined up to 63 elements in wine by ICPMS. Among all atomic spectrometry techniques, ICPMS offers an additional advantage, i.e. the possibility to get isotopic information [70,73–78].

22

G. Grindlay et al. / Analytica Chimica Acta 691 (2011) 18–32

Different ICPMS configurations have been employed in the literature to this end. Sector field [73,75] and time of flight [73,74] mass spectrometers have been mainly used for isotope ratio measurements but quadrupole based instruments have also been employed successfully [33,70,73,76]. 2.3. Interferences The great complexity of the wine matrix usually exerts a negative influence on analytical figures of merit of atomic spectrometry techniques (i.e. accuracy, precision, LODs, etc.). Thus, a good knowledge of matrix effects is required when developing an analytical procedure. In fact, most of the literature about elemental wine analysis by atomic spectrometry techniques deals with matrix effects and how to mitigate them. For this reason, these topics will be thoroughly covered in the next sections. 3. Wine matrix effects Elemental analysis of wine by spectrometric techniques is a challenge due to the high complexity of its matrix (Table 1). Both spectral and non-spectral interferences are commonly related to wine matrix components. In addition, the chemical compounds added to wine during the sample treatment step (mainly acids) must also be considered in order to avoid potential error sources during the determination [79]. As it has been stated above, chemical species in wine matrices can be classified into two main groups: (1) organic compounds: volatile (mainly ethanol) and non-volatile (e.g., organic acids and sugars); and, (2) salts (both inorganic and organic). Fig. 3 summarizes the different potential matrix effects found when analyzing wine samples. Parameters affected by wine matrix components are highlighted by means of dashed lines. As it can be observed in this figure, matrix effects during elemental wine analysis depend on the specific matrix component, on the analytical technique and also on the sample introduction device employed.

Table 2 Median of the volume drop size distribution for primary (pD50 ) and tertiary aerosol (tD50 ) for water, ethanolic solutions and wine. Meinhard nebulizer (type K); nebulizer gas flow 0.7 L min−1 ; sample uptake rate 0.8 mL min−1 . Solution type

pD50

tD50

Water Ethanol 12.5% (v/v) Ethanol 12.5% (v/v) + K 1500 mg L−1 White wine# Red wine#

12.78 11.15 11.15 10.90 10.70

4.46 3.96 3.74 3.76 3.76

#

Ethanol content 12.5% (v/v).

3.1. Organic compounds 3.1.1. ICP-based techniques Ethanol is the most common volatile organic compound present in wines. Its concentration usually ranges between 8% and 19% (v/v). When operating in ICPAES with conventional pneumatic nebulisation of liquids, the presence of ethanol may cause both spectral and non-spectral interferences and degrade analytical performance by introducing noise as it was extensively reviewed by Weir and Blades in the 90’s [80–83]. Regarding non-spectral interferences in ICP-based techniques, firstly it must be considered that the presence of ethanol significantly reduces the solution surface tension and viscosity. As a consequence, aerosols generated by ethanol solutions are finer than water [84]. This effect can be observed in Table 2, which shows the influence of the matrix composition on the median of the volume drop size distribution, D50 , of the primary aerosols (pD50 ) generated using a pneumatic concentric nebulizer and the corresponding tertiary aerosols (tD50 ) introduced into the atomization cell. As it can be observed in this table, ethanol 12.5% (v/v) solution generates aerosols with lower pD50 values (i.e., 12.7% lower) than water, i.e. finer aerosols are obtained for the ethanol solution. As a consequence, ethanol aerosols are more easily transported to the plasma. In addition, the presence of ethanol increases the sample volatility, thus increasing the analyte transport efficiency and reducing

Fig. 3. Summary of the potential interferences occurring during wine analysis by atomic spectrometry methods.

G. Grindlay et al. / Analytica Chimica Acta 691 (2011) 18–32

the median size of the tertiary aerosols, tD50 , (Table 2). As a consequence, signals should be higher than those obtained with water. Nevertheless, the high solvent load into the plasma reduces its energetic characteristics and consequently the signal improvement factor is usually lower than expected [26,35]. In fact, if the amount of solvent reaching the plasma is high enough, the plasma could even be extinguished. Several authors reported on ethanol matrix effects in ICP-based techniques when dealing with wine analysis [32,71,85]. Finally, it must also be taken into account that changes in the sample viscosity could affect the analytical response when operating with free aspiration mode [84]. The presence of carbon in the plasma, irrespective of its source (i.e., volatile or non-volatile organic compounds), causes additional non-spectral interferences [86]. Thus, it has been thoroughly reported that the presence of carbon modifies the signal of some analytes such as As and Se both in ICPAES and ICPMS [30,87,88]. Finally, it is also worth to mention that some carbon deposits on the injection tube and on the skimmer/sampler cones (in ICPMS) may be formed, thus affecting the analytical response [77]. In addition, ICPMS measurements are subject to matrix-induced mass bias effects as reported by several authors when determining Pb [73,74,89], B [70] and Sr [78] isotope ratios in wines. When operating with electrothermal vaporization (ETV) sample introduction devices, some additional matrix effects must be taken into account. As it has been shown in Table 1, the wine matrix contains species such as organic acids (e.g. ascorbic acid and citric acid) which are frequently used as modifiers in ETV [30,90]. Consequently, the signal may be influenced by differences in wine matrix composition. Spectral interferences in ICPAES mainly arise from solvent pyrolysis products such as CN and C2 . They may cause an increase in the background emission spectrum and overlap with some analytical lines. Moreover, intense background signals can degrade analytical figures of merit (mainly LOD) [81,82]. Water can also be a source of spectral interference. Thus, the V most prominent emission line (309.311 nm) suffers from structured background due to vicinity of the OH band at 309.2 nm. Determination of V in wine may therefore not be reliable due to such interference. Accurate V determinations can be obtained at 292.402 nm [26]. In ICPMS, the potential polyatomic interferences due to the presence of carbon compounds must be taken into account (e.g., 40 Ar12 C+ on 52 Cr+ , 12 C16 O+ 28 Si+ , etc.) [35,87]. 3.1.2. Flame atomic absorption spectrometry The sample introduction system mostly used in FAAS is a pneumatic concentric nebulizer coupled to a spray chamber. Due to this fact, the presence of ethanol matrices gives rise to similar non-spectral interferences as previously described for ICP-based techniques. Nevertheless, it is important to highlight that the flame is a more robust medium and thus, it can support a higher solvent load. 3.1.3. Volatile compound generation-based techniques When operating with volatile compound generation-based techniques (i.e., HGAAS, HGAFS and CVAAS), the presence of organic compounds affects the efficiency of the generation of the volatile compound. Wu et al. [91] reported that the HG efficiency for Pb decreased markedly in a medium containing up to 5% (v/v) ethanol in the conventional HGAFS system. Karadjova et al. [92] also found that the As (III) fluorescence signal decreases with increasing ethanol content of the solution. The ethanol interference was attributed to variations in the pH and in the oxidation potentials of substances present in the reaction medium [93]. Finally, it must also be considered that ethanol may interfere with the atomization of hydrides in HGAAS [92]. When dealing with HGAFS, the high sensitivity allows the application of higher dilution factors, thus

23

reducing ethanol levels below 1% (v/v) in the final solution [53], and, therefore diminishing the above mentioned interferences. Capelo et al. [94], also observed that organic compounds present in wine samples affect the reaction of Hg (II) ions with NaBH4 , thus affecting the Hg absorbance signal in CVAAS. Similarly, Segura et al. [52] observed that the amount of NaBH4 required to generate arsine from wine samples was considerably higher than from corresponding aqueous media. Under the selected conditions, significant nonselective background absorption was also observed in HGAAS when AsH3 was directly generated from wine samples. This was attributed to the ethanol vapors transported to the atomizer from the HG system and it was absolutely necessary to use a background correction device [52]. Wu et al. [91] also found that high concentrations of KBH4 cause very high background signals in HGAFS, thus affecting the Pb fluorescence signal. 3.1.4. Electrothermal atomic absorption spectrometry ETAAS analysis of untreated wine samples is difficult and reproducibility is strongly influenced by the presence of ethanol, which affects sample delivery in the furnace [26]. Dessuy et al. [42] also observed that the addition of ethanol to the aqueous standards significantly deteriorated the LOD of Pb. Complex wine matrices may influence the absorbance of some elements and a judicious optimization of the ETAAS experimental conditions and modifier selection must be performed. Thus, Kildahl and Lund [65] found that the use of a matrix modifier is mandatory for the determination of As and Sb in wine samples by ETAAS. Dessuy et al. [42] tested some modifiers for the determination of Pb. These authors found that, opposite to Pd, the addition of ascorbic or citric acids do not significantly affect the Pb or the background signal. At this point, it must be taken into account that non-volatile organic compounds present in wine samples would act as modifier. Kildahl and Lund [65] highlighted the relevance of such compounds in the determination of Cr, Pb and Cd in samples obtained at different stages of the winemaking process. Finally, incomplete pyrolysis of organic matter produces fumes and accumulation of carbonaceous residue after several graphite tube firings which adversely affect the analysis in ETAAS [43,64]. In the analysis of complex samples such as wines, the non-specific background absorption caused by matrix components often causes serious problems [42]. 3.1.5. Total reflection X-ray fluorescence No significant studies about organic matrix effects have been reported in the literature. According to Anjos et al. [48] it is not necessary to apply a correction for matrix effects. 3.2. Salts Wine can be considered as a high salt-content sample (Table 1) containing inorganic anions and easily ionizable elements (EIEs). 3.2.1. ICP-based techniques The influence of EIEs in ICPAES has been extensively reported [95]. Table 2 shows the influence of the addition of K on the median of the volume drop size distribution of the primary aerosols generated by a concentric pneumatic nebulizer. As it can be observed in this table, the presence of K 1500 mg L−1 does not significantly modify the primary D50 values obtained with a 12.5% (v/v) ethanol solution, i.e. similar aerosol characteristics are observed. From Table 2 it is also interesting to note that the tertiary aerosols generated with different wine samples (red and white) are similar to those obtained with the ethanol (12.5% (v/v)) + K (1500 mg L−1 ) solution. As a consequence, no additional non-spectral interferences are expected. Nevertheless, it is also interesting to point

24

G. Grindlay et al. / Analytica Chimica Acta 691 (2011) 18–32

Fig. 4. Sample pretreatment methods employed to mitigate interferences in different atomic spectrometry techniques.

out that the presence of salts modifies the sample volatility thus, affecting the solution transport to the plasma [95]. In addition to the above mentioned effects, spectral interferences have been observed in ICP-based techniques when analyzing wine samples due to interelement effects. Thus, in ICPAES, it has been reported that Cr determination suffers from direct overlap of Fe lines when using the main emission lines (267.716 and 205.552 nm) [26]. Spectral interferences are more frequent when operating with ICPMS. Thus, several authors reported the difficulty of direct measurements of Sr isotope ratios in ICPMS because of the isobaric interference at m/z 87 shown by Rb usually present in wines [75,78]. Other reported interferences are 40 Ca16 O+ on 56 Fe+ ; 39 K16 O+ on 55 Mn; or, 43 Ca16 O+ on 59 Co+ [96]. When operating the ICP with ETV sample introduction, salts can modify the analyte transport to the plasma, thus affecting the analytical response [30]. When working with ICPMS, additional polyatomic spectral interferences can be formed from salts present in the sample. 3.2.2. Flame atomic absorption spectrometry In FAAS, the high concentration of K in wines acts as a natural ionization buffer. Thus, the presence of K in the sample gives rise to enhancements in the absorbance of some alkaline elements such as Na and Rb [26]. It is also worth to mention that anions present in the sample could produce chemical interferences affecting the analytical response in FAAS. PO4 3− are the major anions present in wines (300–800 mg L−1 ). Some elements such as Ba, Ca, Mg, Sr and Al may form refractory PO4 3− thus giving rise to lower absorbance than expected [26]. 3.2.3. Volatile compound generation-based techniques No matrix effects related to the presence of the major components in wine samples (i.e., Na, K, etc.) have been reported using these techniques. Nonetheless, Karadjova et al. [53] reported that transition metals, mainly Cu (II) and Fe (III), are seriously interfering components in wine elemental analysis by HGAAS. Similar issues were observed in HGAFS by Wu et al. [91]. 3.2.4. Electrothermal atomic absorption spectrometry The presence of salts in wine samples can produce matrix effects in ETAAS. Thus, the determination of As is hampered by a highly structured background absorbance probably arising from the high SO4 2− and/or PO4 3− content of some wines [43]. 3.2.5. Total reflection X-ray fluorescence As it has been pointed out previously in Section 3.1.5, no significant studies about matrix effects have been reported.

4. Solving matrix effects in wine elemental analysis Different strategies have been developed in the literature to deal with wine matrix effects. They can get into two groups: general and specific strategies. The former group focuses on the optimization of the sample preparation step as well as the calibration strategy. The main advantage of these methodologies is their universality since they can be employed for the different atomic spectrometry techniques. On the other hand, specific strategies could be employed for a given technique to mitigate matrix effects such as the appropriate selection of the experimental conditions or sample introduction system characteristics. Next, both general and specific methodologies employed in the literature will be briefly discussed. 4.1. General methods 4.1.1. Sample pretreatment methods Fig. 4 shows a diagram of sample preparation methodologies reported in the literature. Sample dilution, conventional dry/wet mineralization and microwave (MW) or ultraviolet (UV) assisted mineralization have mainly been selected to reduce spectral and non-spectral interferences and they represent around 66% of the references to publications on wine analysis by atomic spectrometry. Alternative approaches such as evaporation or extraction are recommended by international organizations or government agencies [5,6] but less frequently reported. Finally, it is important to point out that there are a significant number of papers on direct wine analysis (20%) where interferences are mitigated by selecting the appropriate calibration technique or instrumentation configuration. This topic will be further discussed in the following sections. Table 3 shows an overview of the different sample pretreatments reported in the literature as well as their usefulness to deal with the different wine matrix effects. Additional treatments such as degasification or filtration have also been reported but they depend on wine type or instrumentation characteristics. Thus, for instance, sparkling wine samples require degasification before a wet mineralization step [18,97]. Augagneur et al. [71] reported that wine filtration is required to prevent sample introduction system blockage in ICPMS. 4.1.1.1. Sample dilution. Among the different methods for sample pretreatment proposed to mitigate/reduce wine matrix effects, sample dilution is the only one which simultaneously reduces both inorganic and organic compound related interferences. In addition, sample throughput is almost unaffected since few operation steps are required. Dilution factors ranging from 1:1 to 1:100 (v/v)

[26,28,62,87,96,103] [30,41,65] [30,35,87,89,100,103,112] [85,112] √ √ √ √ √ √ √ √ √ √ √ √ × × × × × × × ×

× × × ×

× × × √ √ Extraction

Evaporation

Matrix matching Standard addition Internal standard Isotopic dilution Calibration

Sample preparation

Dilution Digestion

Water dilution 1:1/1:100 Hot plate Dry mineralization UV-assisted Ozonation High pressure ashing MW-assisted Rotary evaporator Hot plate Infrared radiation Liquid–liquid extraction Solid phase extraction

Non-volatile √ √ √ √ √ √ √

Non-volatile √ √ √ √ √ √ √

× × × × × × × × × √ √ × × × √ √

× × × × × × × × × √ √

√

Volatile √ √ √ √ √ √ √ √ √ √ √ √ Volatile √ √ √ √ √ √ √ √ √ √ √ √

√

Spectral Non-spectral Spectral

Non-spectral Salts Organic compounds

Mitigated interferences Methodology

Table 3 √ Advantages and drawbacks of the general methods employed to mitigate/eliminate wine matrix effects: ( ) matrix effects successfully mitigated; (×) matrix effects not mitigated.

References

[35,37,53,70,87,88,96,99–101] [13,37,42,52,62,64,79,85] [36,52,101,102] [28,29,77,78,89] [94] [72] [35,37–39,70,77,103] [101,102,104] [36,105] [67,106] [54,74,107] [55–58,108]

G. Grindlay et al. / Analytica Chimica Acta 691 (2011) 18–32

25

[35,37,53,70,87,88,96,98–102] have been employed but the appropriate dilution factor depends on the method of analysis as well as on the wine composition and analyte concentration. It is obvious that using a high dilution factor decreases the magnitude of matrix effects. However, a high dilution factor could hamper the determination of trace or ultratrace elements. Thus, 1:1 up to 1:5 (v/v) dilution is usually employed for AAS techniques [53,64] whereas the high detection capabilities of ICPMS allow 1:10 (v/v) or even a higher dilution factor [37,68,70,100]. Grindlay et al. [35] have studied sample dilution methodology for elemental analysis of wine using ICP-based techniques. Matrix effects were almost eliminated when a 1:5 dilution factor was employed for ICPAES analysis. However, strong signal suppression in ICPMS was still present due to EIEs even after 1:10 (v/v) dilution. It is important to point out that dilution may affect wine pH and hence, solution stability [35]. K and Ca tartrate salts are not soluble at high pH values and they tend to precipitate if no acid is added [2,3]. 4.1.1.2. Sample digestion. Sample digestion methods have been used successfully for elemental wine analysis since wine organic components are efficiently decomposed, and matrix effects are then reduced [13,36,52,85]. Nonetheless, it is important to keep in mind that additional matrix effects can arise from the use of chemicals to decompose the wine matrix [79]. Wine is usually heated on a hot plate in a polytetrafluorethylene (PTFE) vessel with HNO3 [85] and/or additional reagents, e.g. H2 O2 [42,52], H2 SO4 [52], HClO4 [13] or V2 O5 [64]. Dry mineralization (with Mg(NO3 )2 + MgO) has also been employed successfully [36,52,101,102]. The main drawbacks of these treatments are: (i) the high amount of reagents required (ii) the low analysis throughput since the time required may be up to 24 h [85]; (iii) potential losses of volatile elements [37,62]; and (iv) contamination problems due to reagent impurities. In order to solve these issues, alternative digestion methodologies have been proposed in the literature. Thus, for instance, O3 assisted digestion procedures [94] easily decompose wine samples at ambient temperature in less than 1 h. In addition, the amount of reagents required is drastically reduced (i.e. lower contamination risks). UV-assisted digestion treatment has been extensively employed by Almeida et al. [28,29,77,78,89]. A sample of 20 mL of wine is spiked with 120 ␮L 30% (w/w) hydrogen peroxide and the mixture is irradiated by UV radiation from a 1000 W mercury high pressure vapor lamp. The free OH radicals generated by H2 O2 and UV radiation digest the wine matrix in less than 2 h. Nonetheless, a post-digestion filtration step could be required due to uncompleted sample decomposition. Finally, the efficiency and speed of wine decomposition can also be improved with the aid of MW radiation [35,37–39,70,77,103]. The sample–acid mixture is exposed to a high pressure and temperature treatment in a closed PTFE vessel irradiated by MW. As a consequence of these extreme conditions, less time and reagents are required to destroy the wine matrix when compared to the conventional digestion procedure. In fact, complex acid mixtures (i.e. HClO4 , H2 SO4 , etc.) are not required and HNO3 [35,38] or HNO3 + H2 O2 [37,39,70,77,103] are usually enough to decompose the wine matrix. Gonzálvez et al. [37] compared different sample preparation strategies for simultaneous multielemental analysis of wine using ICP-based techniques and they concluded that MW-assisted digestion should be preferred over thermal digestion treatment since analyte losses are prevented and no additional reagents such as HClO4 are required. In MW-assisted digestion, it must be taken into account that once the acids are added to the vessels, samples should be kept in the fume hood until NOx gases are removed to avoid that sample pressure could provoke vessels venting [35]. As a consequence, sample throughput is negatively affected. In addition, the digestion of sugar rich wines may lead to the formation of a carbonaceous

26

G. Grindlay et al. / Analytica Chimica Acta 691 (2011) 18–32

residue inside MW vessels, making necessary the use of softer conditions.

(w/w) + 1000 mg L−1 K were able to match the sensitivity obtained for wine samples [30].

4.1.1.3. Other methods of sample pretreatment. Less common sample preparation approaches in the literature include dealcoholisation [36,101–106] or analyte separation. Alcohol can be removed using a rotary evaporator [101,102,104], by heating the sample either on a hot plate [36,105] or by means of IR radiation [106]. Once the alcohol is removed, the sample volume is adjusted with deionized water or HNO3 . A comparative study of dry mineralization, wet mineralization and dealcoholisation showed that all three methods lead to unbiased results [36]. However, Olalla et al. [101,102] reported that wet digestion (i.e. HNO3 + H2 SO4 ) provides the best analytical figures of merit (i.e. sensitivity, linearility, LOD, precision) when compared to dealcoholisation or dry mineralization for Ca and Fe determination by FAAS. Other authors preferred to separate analyte from the matrix by liquid–liquid extraction [54,74,107] or SPE [55–58,108] to reduce matrix effects and to preconcentrate the analyte. In fact, preconcentration is widely used with FAAS techniques. Nonetheless, this approach is time consuming because column elution and previous digestion steps are required [56–58,108].

4.1.2.3. Internal standardization. The internal standardization methodology has also been employed to mitigate matrix effects as well as to obtain accurate results and good repeatability. The success of this methodology depends on the selection of the internal standard element(s) and requires simultaneous measurement of the analytical and reference signals. For this reason, internal standardization has been mainly employed for ICP-based techniques although Ferreira et al. reported that it could be implemented with fast sequential multielement FAAS [45]. In order to be successful in ICPMS, both the internal standard and the analyte must have similar mass and behave similarly in the plasma [110,111]. Thus, for instance, Tl (m/z: 205 I.P: 6.1 eV) has been employed as internal standard for Pb (m/z: 208 I.P: 7.4 eV) determination in wine [87,89,112]. In ICPMS, Roudskin et al. [87] pointed out that As and Se determination could not be performed using In as internal standard since In is not affected by intensity changes related to the carbon charge transfer mechanism, like As and Se. For this reason, Au has been used as an internal standard element in the determination of As and Se because it is affected by the same interference [30]. When multielement analysis has to be carried out, it is not easy to find an internal standard which fulfils previous requirements for all analytes of interest. Iglesias et al. [103] employed two internal standards for the multielemental analysis of wine by means of ICPAES and ICPMS. Ni and Y emission lines were used in the determination of 11 elements in wine by ICPAES whereas 103 Rh was used as internal standard in ICPMS for m/z between 7 and 121 amu and 205 Tl for 137–208 amu [103]. Nonetheless, several authors have claimed that one single internal standard (usually 115 In+ or 103 Rh+ ) is enough for successful multielemental analysis of wines [32,100,113]. The appropriate selection of the internal standard for multielemental analysis of wine in ICP-based techniques has been studied in more detail by Grindlay et al. [35]. The authors evaluated Sc, Ti and Be as internal standards for ICPAES and 7 Be+ , 45 Sc+ , 140 Ce+ and 238 U+ as internal standards for ICPMS. The Ar 396.152 nm emission line was also evaluated for internal standardization since its intensity was affected by wine matrix components. Internal standard selection depends on the analyte(s) as well as the matrix and sample introduction system. Matrix effects have been corrected by using Sc in both ICPAES and ICPMS but special attention should be paid to line characteristics in optical emission. Thus, Sc II 361.384 nm emission line is unable to correct matrix effects whereas the Sc II 357.253 nm emission line can be used successfully [35].

4.1.2. Calibration strategies The selection of the calibration strategy for elemental analysis of wine by atomic spectrometry techniques is closely linked to the sample preparation step and experimental conditions. Calibration strategies employed in the literature are: (i) matrix matching; (ii) standard addition; (iii) internal standardization and (iv) isotope dilution when operating ICPMS. 4.1.2.1. Matrix matching. Most of the work published on elemental analysis of wine is based on external calibration with matrix matched standards, although exact simulation of the wine matrix is difficult due to its complexity. However, it has been shown previously that matrix effects mainly arise from the presence of ethanol and EIEs. Thus, when samples are analyzed directly or after dilution, the appropriate amount of ethanol is added to the standards [28,87,96,103]. Only for the analysis of major elements (i.e. Na, Ca or Mg) by FAAS, an ionization buffer such as K is added to the samples [26,62]. When digestion or extraction procedures are used, the acids or reagents employed to decompose or extract wine are also added to the standards [28,35,103]. Despite reported ethanol and EIEs related matrix effects, several authors have demonstrated that external calibration is possible using aqueous standards [35,53,64,109]. Unfortunately, it is not always clear how “aqueous standards” are exactly defined in the literature (i.e. only water or acid containing solutions, etc.). According to dos Santos et al. [109] no differences in the slopes of calibration curves for aqueous and acid containing standards in FAAS were observed. Similar results have been reported for HGAFS [53] and ETAAS [64]. The application of microwave based desolvation systems to remove ethanol from the aerosol stream also allows calibration with aqueous standards in ICPAES [35]. 4.1.2.2. Standard addition. Several authors have reported on elemental analysis of wine where aqueous or matrix matched standards could not provide unbiased results and alternative methodologies such as standard additions were required to compensate for matrix effects [30,41,65]. Thus, for instance, Kildahl and Lund [65] were forced to use standard addition due to the differences in the As and Sb sensitivity between standards and untreated and decomposed wine samples in ETAAS. Similar issues were reported in ETV-ICPMS where none of the matrix matched standards prepared in e.g. ethanol 10% (w/w) or ethanol 10%

4.1.2.4. Isotopic dilution. This calibration methodology can be employed in ICPMS due to the possibility to get isotopic information. Isotopic dilution involves the addition of a known amount of an enriched isotope of the element of interest to the wine sample. By measuring the isotope ratio of the spiked sample and knowing the isotopic ratio of the spike, the analyte concentration is calculated. Analytical results based on isotope dilution are not influenced by instrument drift or by non-spectroscopic interferences, although special attention should be paid to spectral interferences and contamination. In addition, preliminary semi-quantitative analysis may be required to determine the right amount of the spike to be added [112]. Up to now, despite benefits, this methodology has only been applied to Pb determination in wine [33,73,85,112]. 4.2. Specific methods Despite reported problems caused by the wine matrix composition, direct wine analysis is sometimes feasible by the appropriate

G. Grindlay et al. / Analytica Chimica Acta 691 (2011) 18–32

27

Table 4 √ Summary of specific methods to deal with wine matrix effects: ( ) matrix effects successfully mitigated; (×) matrix effects not mitigated. Technique

Methodology

Mitigated interferences

References

Organic compounds Spectral

ICPAES ICPMS

Plasma robust conditions Desolvation system Mathematical equations Sector field ICPMS Plasma robust conditions Micronebulizer Flow injection analysis Desolvation system Electrothermal vaporization

HGAAS/HGAFS/CVAAS

Reaction conditions Flow injection analysis Anti-foaming agents

ETAAS

Furnace type Temperature program Modifiers

Volatile √ √

Non-volatile √

√ √ √

× √ √ √

× × √ √

× × × √

√ √ √

√ √ √

× √ √

× √ √

selection of the experimental conditions or instrument characteristics (i.e. instrument design, sample introduction system, etc.). Such specific applications will be discussed in this section. Table 4 shows an overview of the different approaches reported in the literature as well as their usefulness to deal with the different wine matrix effects. 4.2.1. ICP-based techniques When working with ICP-based techniques, robust conditions (i.e. low nebulizer flow rate and high r.f. power values) have been recommended to improve analyte atomization and ionization as well as to reduce spectral and non-spectral interferences [35,100]. Optimum plasma conditions are of special importance in ICPMS, where incomplete aerosol atomization increases oxide levels and matrix related spectral interferences [100]. Nonetheless, Bianchi et al. [69] reported that cool plasma conditions (i.e. non-robust conditions) reduce high background signals for light isotopes and plasma based spectral interferences (i.e. 40 Ar16 O+ ) enabling Li, Al and Fe determination in wine by ICPMS. However, these authors did not discuss matrix based interferences (e.g. 40 Ca16 O+ ) which are expected to be significant. Spectral interferences due to matrix components in ICPMS can be partially mitigated by the application of correction equations [28,29]. Thus, for instance, Ca based interferences (40 Ca16 O1 H, 43 Ca16 O, 44 Ca16 O) over 57 Fe, 59 Co and 60 Ni isotopes were corrected by measuring the Ca signal at 40, 43 and 44 m/z. This methodology has also been applied to mitigate 40 Ar35 Cl and 135 Ba16 O spectral interferences on 75 As and 151 Eu determination [28,29]. It should be noted that such correction equations are also required for isotopic ratio measurements [77,78]. Although wine does not contain significant amounts of Hg, the 204 Pb signal should be corrected to take into account a contribution of 204 Hg [77]. A similar procedure has been employed to correct for 87 Rb contribution on the 87 Sr signal [78]. Spectral interferences in ICPMS can also be addressed by appropriate selection of instrument configuration. Thus, sector field ICPMS allows to mitigate the spectral interferences observed with quadrupole ICPMS due to its better resolution capabilities [87,96]. For instance, medium resolution eliminates typical interference in wine analysis such as 40 Ca16 O+ /40 Ar16 O+ on 56 Fe+ , 39 K16 O+ on 55 Mn+ or 40 Ar12 C+ on 52 Cr+ determination. The selection of the sample introduction system also plays an important role in the magnitude of matrix effects. The conventional sample introduction system in ICP techniques (i.e. a pneumatic neb-

Salts Non-spectral

Spectral

Non-spectral

Volatile √ √

Non-volatile √

√

√ ×

× × √

× × √ √ √

× √ √ √

× × √ √ × × × × √ √

×

× × × × × √ × √

× × × √ √ √ √ × √ √

× × √ √ √ × × × × × √ × √

[35,110] [35] [28,29,77,79] [87,96] [98] [71] [88,112] [32,35] [30] [52] [53] [51] [26,42,64] [39,42,43,64] [39,41,43,64]

ulizer attached to a spray chamber) shows several drawbacks which make it difficult to determine trace and ultratrace elements in complex samples, such as: (i) clogging problems with high dissolved solid concentration; (ii) low analyte transport efficiency; and (iii) high solvent and matrix plasma load when operating at optimum liquid sample uptake rate (≈1 mL min−1 ). Several authors used high efficient sample introduction systems to improve the analyte transport rate meanwhile decreasing the amount of sample consumed. Augagneur et al. [71] reported that matrix effects were virtually absent when using a micro-concentric nebulizer whereas for a Vgroove nebulizer, the signal is suppressed by a factor of 2–5. The difference in the magnitude of matrix effects observed with both systems is related to their operating conditions. Micro-concentric nebulizers have been designed to operate at low sample uptake rates (<200 ␮L min−1 ) and as a consequence, the amount of matrix transported to the plasma is very low (i.e. lower matrix effects). Matrix effects with the V-groove nebulizer could also be mitigated by reducing the Ql but the sensitivity would decrease accordingly. Reduction of matrix effects can also be easily accomplished by coupling a FI system to the nebulizer [88,112]. FI systems introduce a discrete amount of sample without affecting analytical figures of merit significantly but decreasing the amount of sample (and therefore also matrix) delivered into the plasma [112]. Thus, for instance, when a FI system was employed, no difference in As response was observed between ethanol and pure aqueous solutions in ICPMS [88]. On the other hand, similar experiments with a conventional sample introduction system lead to a decrease in As signals (40%) as well as optimum nebulizer flow rate for a 10% (v/v) ethanol solution. According to Wangkarn and Pergantis [88] FI figures of merit in ICPMS could be further improved using a micro-FI system. By means of this approach, carbon based non-spectral interferences on As signals are reduced 2–3 times when compared with a conventional FI system. Desolvation systems are useful to reduce ethanol based matrix effects [32,35]. These devices are especially useful to mitigate wine ethanol matrix effects in the plasma. Thus, for instance, no differences in the sensitivity and LOD’s are observed in ICPAES for a microwave based desolvation system when operating with pure aqueous solution or ethanol 10% (v/v). Consequently, wine can be directly analyzed without sample preparation using aqueous standards [35]. Although this desolvation system is not able to reduce completely the non-spectral matrix effects of ethanol in ICPMS, the background at m/z 52 (from 40 Ar12 C+ ), interfering the 52 Cr

28

G. Grindlay et al. / Analytica Chimica Acta 691 (2011) 18–32

determination, is 10 times lower than for a conventional pneumatic nebulizer attached to a double pass spray chamber [35]. Desolvation systems reduce organic compound based interferences but they cannot reduce interferences due to the presence of EIEs or non-volatile organic components. In fact, those interferences are enhanced in comparison to a conventional sample introduction system [32,35]. Thus, for instance, analyte recovery assays for nondiluted wine samples in ICPAES have shown that recovery values were on average 50% lower than expected due to EIE matrix effects when a desolvation system was used and 70% without desolvation. Matrix effects in ICPMS operating a desolvation system were even more significant since, after 1:10 (v/v) dilution, recovery values were still on average 50% lower than expected [35]. Electrothermal vaporization devices are a useful alternative to conventional sample introduction systems due to their capability to analyze wine directly and remove wine matrix components (i.e. mainly ethanol) through a temperature program [30]. ETV was successful to reduce 40 Ar35 Cl+ spectral interference on 75 As+ when compared to a conventional sample introduction system in ICPMS. However, ETV does not mitigate interferences due to the presence of non-volatile organic compounds and EIEs [30]. When the wine volume analyzed is higher than 4 ␮L, signals are depressed due to changes in vaporization and transport to the plasma [30]. Finally, it is important to note that for all sample introduction systems or experimental conditions employed, K and organic matter present in wine can cause blockage of the injector tube and cones due to incomplete pyrolysis of wine matrix within the plasma, affecting long-term ICPMS performance. These problems may be prevented by means of an extra oxygen stream to the plasma region [114] but special attention should then be paid to the increase of oxygen based spectral interferences. 4.2.2. Flame atomic absorption spectrometry Studies on how to mitigate matrix effects by the selection of the experimental conditions or instrument characteristics in FAAS have not been recently found, probably due to the flame robustness against wine matrix components. Nonetheless, when dealing with matrix effects due to refractory compounds, matrix effects are easily mitigated by optimizing gas flows. Huang et al. [44] reported that sulphur determination by continuum source FAAS via molecular absorption of carbon monosulfide requires rich fuel conditions to avoid that carbon monosulfide reacts with oxygen. 4.2.3. Volatile compound generation-based techniques Selection of optimum experimental conditions to reduce matrix effects with HG based techniques (HGAAS or HGAFS) is challenging due to the great amount of variables involved. According to T˘asev et al. [67], As determination in untreated wine samples by HGAAS was not possible due to strong ethanol matrix effects and, as a consequence, evaporation pretreatment to remove ethanol was mandatory. Segura et al. [52] reported that direct wine analysis for As determination was not feasible by HGAAS despite increasing the amount of NaBH4 employed for ethanol solutions in comparison with aqueous media. Similar observations were made in HGAFS where the analytical signal did not reach maximum value when experimental parameters such as HCl or NaBH4 concentration and flow rate were optimized within a wide range of experimental values [52]. Sample dilution was mandatory to mitigate matrix effects [53,92]. Nonetheless, several authors have pointed out that direct wine analysis is still feasible but FI devices are required [51–53]. Optimum experimental conditions for direct wine analysis are extremely linked to the chemical form of the analyte. When Pb analysis is carried out by FI-HGAAS, plumbane formation is favoured by using K3 Fe(CN)6 as pre-oxidation agent to transform Pb (II) into Pb (IV) [51]. Pb signal was mainly controlled by the K3 Fe(CN)6 flow rate as well as HCl concentration whereas little effect of reaction

coil, argon carrier flow rate and both NaBH4 concentration and flow rate were observed. Segura et al. [52] also reported that the chemical form of As has a high impact on analytical results. Thus, for instance, both inorganic As species (As (III) and As (V)) can be determined together by using 9 mol L−1 HCl and 0.2% (w/v) NaBH4 after pre-reduction of As (V) with KI, otherwise only As (III) is determined [67]. These experimental conditions do not allow generation of arsine from organic As compounds and a digestion treatment is required to quantify them from the difference between total As and inorganic As content. Karadjova et al. pointed out that both inorganic and organic As species could be determined simultaneously using l-cysteine as a pre-reduction and complexation agent [53]. The use of different acids such as citric acid, acetic acid or HCl allows selective determination of inorganic and organic As. High background signals originating from the organic components present in wine were mitigated by using a highly oxidizing flame [51]. On the other hand, when working with hydrogen diffusion flame – AFS, analytical figures of merit are improved by replacing the argon carrier gas by hydrogen [52]. Thus, hydrogen flame stability is less dependent on the HG process and less concentrated reagents are required for wine analysis. Finally, it is important to point out that, for wine rich in organic compounds, anti-foaming agents are required to suppress bubble formation in the liquid–gas separator and improve analyte transport efficiency. However, the concentration should be carefully optimized since the analyte signal could be depressed [51,91]. 4.2.4. Electrothermal atomic absorption spectrometry Furnace characteristics, temperature program and modifier selection are the key parameters to reduce matrix effects as well as to control analytical figures of merit in ETAAS. As expected from previous ETAAS fundamental studies, stabilized temperature platform and transversally heated furnace conditions are preferred for elemental wine analysis in order to ensure complete analyte atomization and avoid temperature gradients inside the graphite tube which may cause matrix effects [26,64]. The influence of the furnace design on Pb determination in wine has been evaluated by Dessuy et al. [42]. These authors observed that a transversally heated filter furnace affords better precision and higher signal-tobackground ratios than a transversally heated graphite tube with integrated platform. It is interesting to point out that the ETAAS work in this field is usually performed by means of Zeeman-effect background correction based instruments due to their capability to deal correctly with high and complex background absorbance. The development of the furnace temperature program for ETAAS analysis is not an easy task since wine affects sample delivery as well as optimum drying and pyrolysis temperature. Thus, in order to prevent sample sputtering and/or foaming of wine samples with a high organic content, the furnace is usually pre-heated for sample deposition [39,42,43]. According to Soares et al. [115], sample delivery could be improved with the aid of Triton X-100. On the other hand, Ajtony et al. [43] have pointed out that the ramp time and the length of the drying step should be carefully optimized. These authors employed a two-step drying procedure at 110 and 130 ◦ C for the analysis of must and wine samples. Even a triple drying step has been employed [42]. Optimum pyrolysis and atomization temperature for a given analyte are closely related to both modifier and wine matrix characteristics. Thus, a two-step pyrolysis at 400 and 700 ◦ C was employed to improve long-term instrument performance (i.e. precision, accuracy, etc.) [64]. An extended pyrolysis step is useful to reduce fume generation during pyrolysis and avoid the build up of carbonaceous residues in the furnace after several firings. Extending the lifetime of the graphite tube and improving matrix removal is also possible by addition of HNO3 or H2 O2 [39,43]. Kristl et al. [39] have shown that, when 10% (v/v) HNO3 is employed

G. Grindlay et al. / Analytica Chimica Acta 691 (2011) 18–32

to oxidize the wine matrix components, only a one-step pyrolysis program is required. Matrix removal is improved by increasing the pyrolysis temperature and a modifier is usually required to stabilize the analyte. Thus, when Pd(NO3 )2 /Mg mixture is employed as modifier, the maximum pyrolysis temperature for Cd and Pb determination in white and red wines was enhanced from 200 to 600 ◦ C and 1000 ◦ C, respectively [64]. Both Pd(NO3 )2 or Pd(NO3 )2 /Mg have been widely used as universal modifiers for As, Pb, Cd and Cu determination in wine [39,43]. A mixture of (NH4 )3 PO4 /Mg(NO3 )2 provides similar results as obtained with Pd-based modifiers but the later are preferred to avoid structured background problems caused by PO4 3− [39,64]. Several authors have evaluated permanent modifiers for elemental wine analysis [41–43]. In general, non-permanent modifiers offer better analytical figures of merit than permanent ones. Thus, for instance, Pb, Cu and Cd sensitivity obtained with Ir as permanent modifier was lower than when the Pd(NO3 )2 or Pd(NO3 )2 /Mg modifier was used [43]. In addition, Dessuy et al. [42] reported that the Ir modifier leads to a pronounced double peak for Pb, similar to that obtained without using a modifier, which indicates stabilization problems. According to Karadjova et al. [41], only pre-reduced Pd modifier was able to simultaneously stabilize the different Hg organometalic species studied (i.e. methylmercury (II) chloride, dimethyl mercury, dyethylmercury and dyphenylmercury). Finally, organic modifiers have not been extensively evaluated for elemental wine analysis in ETAAS. As shown in Table 1, wine has significant amounts of different organic acids (i.e. ascorbic acid, citric acid, etc.) and it is not an easy task to determine the additional amount of organic modifier required for a given sample. Thus, for instance, ascorbic acid or citric acid did not improve analyte stabilization and reduce the background absorbance when compared to Pd modifier for Pb determination in Brazilian, Chilean and Spanish wines [42]. 5. Results assessment Before the analysis of unknown wine sample can be performed, it is important to check whether the analytical methodology developed is accurate, precise and free from systematic errors. To this end, different approaches have been employed in the literature such as: (i) the analysis of wine reference materials; (ii) comparison with additional methods/techniques; and (iii) recovery studies.

29

methods in the literature are negligible. The reference methods are sometimes confusing and no universal sample pretreatment or calibration strategy is included for all the analytes regulated. Thus, for instance, Fe and Zn determination according to European Union official methods [6] is performed after ethanol removal by evaporation. However, “an appropriate dilution” is recommended for Cu determination, although the concentration level for this element is quite similar to that found for Fe and Zn. On the other hand, Ag determination is recommended by FAAS after a digestion treatment but the Ag concentration level in wine is far below the LOD of the FAAS technique [29,118]. Results assessment for a certain wine analysis method can also be evaluated by comparison with alternative sample preparation treatments or calibration strategies [35–37]. For instance, Segura et al. [52] evaluated the accuracy of As determination in un-diluted wine samples by FI-HGAFS by comparing the results for different mineralization procedures. For example, Goosens et al. employed standard addition and isotope dilution methods for the determination of Pb [85]. In ICP-based techniques, analysis after digestion is commonly employed to check accuracy and matrix effects for direct wine analysis [35,85]. Finally, different instrument configurations or analytical techniques have also been used for results assessment. Thus, for instance, sector field, quadrupole and time of flight based ICPMS were employed to check accuracy and precision of lead isotope ratios in wine [74]. Rodushkin et al. employed thermal ionization mass spectrometry to evaluate the results obtained with sector field ICPMS [87]. Results obtained with ICPMS and ETAAS have been employed to the same samples to check the accuracy [30,89]. ICPMS results have also been compared successfully with those obtained by TXRF [32] or neutron activation analysis [68]. 5.3. Recovery studies The accuracy and the precision of wine analysis methodology have traditionally been evaluated by means of recovery studies when certified materials or an alternative techniques are not available [25,35]. Although this approach is easy to implement, special attention should be paid when working with volatile compound generation-based techniques (i.e. HGAAS, HGAFS, CVAAS) since analyte oxidation state exerts a great influence on the analyte signal.

5.1. Wine reference materials

6. Speciation

Over the years, several certified metal wine samples have been developed by the Institute for Reference Materials and Measurements (IRMM) from the Community Bureau of Reference (Brussels, Belgium) as part of different inter-laboratory comparison programmes [33]. These materials (i.e. BCR B, BCR C, BCR E and IMEP16) provided certified lead concentration values but their availability was very limited and nowadays they are no longer available. Recently, commercial certified wine samples have become available in the market [116,117]. Thus, the Chambre d’Agriculture de la Gironde (Blanquefort, France) provides reference wines with certified concentrations of only major elements such as K, Ca, Cu and Fe [116]. Cd and Pb certified wine samples have been developed by The Food and Environment Research Agency (USA) [117]. It is interesting to point out that, despite the complexity and the unstable nature of wine, these certified materials are warranted up to 5 years.

Speciation analysis of wine samples is receiving increasing attention because the additional information on the identity and quantity of species of trace elements is important in the assessment of toxicity. Bioavailability, mobility and toxicity depend on the specific chemical form of the species. Metals may exist in wines as free ions, as complexes with organic acids or as large molecules, e.g. polysaccharides, peptides, proteins and polyphenols. Reviews on chemical speciation and fractionation of metals in wine have ´ been published by Pyrzynska in 2004 [25] and in 2007 [119]. McKinnon and Scollary studied the size fractionation of metals in wines and found cations (K, Na), metal tartrate complexes (Al, Cu, Ca, Fe) and PO4 3− complexes (Fe) using ultrafiltration techniques combined with elemental analysis [120]. The binding agents for Pb were residual proteins and procyanidines in white wine and tannins in red wine. Fractionation patterns for Pb in red wine showed a decrease between 100,000 and 30,000 nominal molecular weight, indicating the binding with tannins [120]. Size fractionation of nonvolatile dissolved organic compounds and metal species has been achieved by multistage ultrafiltration coupled to ICPMS, leading to the conclusion that, e.g., most of the Ba and Sr species were

5.2. Methods comparison Different government agencies [5,6] have published official methods for wine analysis but reported applications of reference

30

G. Grindlay et al. / Analytica Chimica Acta 691 (2011) 18–32

found in the lowest molecular fractions (<1 kDa) and most of the Pb species in higher molecular fractions (>1 kDa) [121]. Murányi and Papp used filters of different pore size to investigate which fraction of a particular metal (Cd, Pb, Cu, Zn, Mn, Al, Fe, Ca) can be found in the solution, in colloidal form or in the form of suspension and concluded that the reduction of a specific metal due to filtering shows considerable differences related to former treatment and storage of the wine [122]. Complexation with biological ligands is also receiving attention. Pérez-Corona et al. studied for example the biotransformation of inorganic Se added to grape juice in the fermentation process of wine and found that selenomethionine was the main Se species formed [123]. It is important to note that the metal species should not change during the sampling, storage and analysis of wines, i.e. every step in the analytical process should be carefully considered in the evaluation and validation of speciation analysis. It should be noted that international legislation on trace elements as contaminants in wine is based on total element concentrations [124]. So far, most studies have focused on speciation of Pb, As, Fe and Cu in wine, but also Zn, Sn and Mn received some attention. Speciation includes elucidation of the oxidation state, but it also includes the chemical form of species. Thus, for instance, when dealing with As determination in wines, quantification of arsenite (As (III)) and arsenate (As (V)) is required, but also additional information on molecular As species is of interest, e.g. monomethyl arsenic acid (MMA) and dimethylarsinic acid (DMA), because all species differ in biological properties and toxicity [125]. Hyphenated techniques are generally used for speciation analysis because they offer separation of the species in combination with high sensitivity atomic spectrometry detection of the elements of interest [124]. For wine analysis, the coupling of high performance liquid chromatography (HPLC) with ICPMS or ICPAES is the most attractive approach, although other methods have also been reported [126]. Good separation of the species of interest often requires elution with solutions that cause problems for stable operation of the plasma, especially when gradient elution is applied. However, the amount of solution that reaches the plasma can be reduced by using micronebulizers and further reduction is achieved by cooling of the spray chamber. Some applications of plasma spectrometry for elemental speciation in beverages have been reviewed by Meija et al. [126] in 2004. It is important to point out that when speciation studies are carried out, alternative techniques (i.e. molecular absorption spectrometry, electrochemical methods, etc.) [127–130] could be required to study successfully metal complexes in wine.

6.1. Pb speciation Lead speciation in wines is a classic issue. Thus, Lobinski et al. [131] reported that organolead compounds were found in vintages of renowned French wines collected in the same area over 40 years from an area of intensive automotive pollution. The authors observed that the pattern of trimethyllead variation with time followed the consumption of leaded gasoline in Western Europe and also concluded that natural biomethylation of inorganic Pb does not occur in wine. Teissedre et al. confirmed that no methylation of Pb was observed while slow degradation of triethyllead occurred during the fermentation process [132]. Szpunar et al. found that Pb was mostly (40–95%) bound as a complex with the dimer of a pectic polysaccharide using size exclusion HPLC with on-line detection of Pb by ICPMS [104]. Further information on the speciation and bioavailability of Pb has been obtained by electrochemical methods of analysis, e.g. by Azenha and Vasconcelos, who found that the dialyzable fraction of Pb during intestinal digestion was rather low for all of the studied wines [133,134].

6.2. As speciation Chromatographic as well as non-chromatographic procedures have been used for speciation of As in wine. Karadjova et al. used continuous flow HGAFS for determination of total As and they could also selectively measure the concentrations of arsenite, arsenate, MMA and DMA by adjusting the experimental conditions for arsine generation [92]. For comparison, As species in wine were also determined by coupling of ion chromatographic separation with HG-flame AFS detection. For the wines studied, arsenite was found as major As species. Detection limits were around 0.1–0.4 ␮g L−1 using both methods [41]. Taˇsev et al. reported on the determination of inorganic As species and total As in wines by means of HGAAS and they also found that arsenite was the predominant As species in wine [67]. In the more than 50 wine samples analyzed, no organic As species were found. Herce-Pagliai et al. determined total As and inorganic and organic As species in 45 wine samples by HGAAS and ion-exchange chromatography coupled to HGAAS. They found As levels ranging from 2 to 15 ␮g L−1 where DMA was the most abundant species in most of the samples [135]. However, the concentrations were relatively low considering international legislation and the estimated daily intake of total As for consumers [135]. Wangkarn and Pergantis developed a method for the high-speed ion-pair reversed-phase narrow-bore HPLC separation of inorganic As, MMA and DMA, on-line coupled to ICPMS and found only trace levels of arsenite in the wine samples. Detection limits were at the low ␮g L−1 As level [136]. In a recent study, Moreira et al. used HPLC coupled to ICPMS for As speciation in white wines produced in South America and found 3–10 ␮g L−1 As (III), 9–18 ␮g L− 1 As (V) and up to 1.1 ␮g L−1 DMA [137]. An alternative method for speciation of As is capillary gas chromatography with AES. This approach requires the formation of derivatives, e.g. with methyl thioglycolate and allows the determination of inorganic As, MMA and DMA [124]. 6.3. Fe speciation Fractionation and speciation studies of Fe in wine samples have been reported by Karadjova et al. [57], using FAAS and ETAAS for the determination of Fe in different fractions. In bottled wine almost 20% of the organically bonded Fe was found in complexes with polyphenols and proteins and less than 5% was found in the polysaccharide fraction, depending on the type of wine and the wine processing. The concentration of labile Fe (II) is usually higher than the concentration of labile Fe (III) [57]. Tawali and Schwedt used a combination of SPE and FAAS for the differentiated analysis of labile Fe (II) and Fe (III) species in wines and compared the results with those obtained with adsorptive stripping voltammetry [138]. Paleologos et al. developed a micelle mediated methodology for the determination of free and bound Fe in wines by FAAS based on precipitation of tannins and other phenolic and insoluble compounds in the micelles of a non-ionic surfactant mixture [139]. The authors found that the amount of Fe existing as an insoluble complex with tannins and other related compounds varied from 5% in commercially available white wines to 30% in domestic red wines. de Campos Costa and Araújo used a sequential injection analysis system with FAAS for the sequential determination of Fe (III) and total Fe in wines, based on the extraction with MIBK of the complex formed between Fe (III) and thiocyanate [54]. Ionexchange chromatography coupled to FAAS was used by Ajlec and ˇ Stupar for determination of the ratio of Fe (II) and Fe (III) in wines [140]. Tasev et al. used liquid/liquid extraction and column SPE procedures combined with FAAS for Fe species determination in wines from Bulgaria and Macedonia and found <15% Fe (II), 30–40% Fe (III) and 30–50% organically stable bounded Fe species [141]. A tandem column SPE and FAAS were used by Pohl and Prusisz

G. Grindlay et al. / Analytica Chimica Acta 691 (2011) 18–32

for the determination of inorganic and organically bound forms of Fe in wine [142]. Three different Fe groupings were discriminated and assessed including hydrophobic species of Fe bound to phenolic substances and related species, cationic species comprising Fe ions and labile Fe forms, in addition to anionic and/or neutral Fe complexes with organic acids. They found the method very useful for examining the chemistry of wine oxidation [142]. Other approaches for speciation of Fe were based on molecular absorption spectrometry [127], fluorescence spectrometry [143] or electrochemical detection in combination with FAAS [128]. 6.4. Cu speciation In their evaluation of Cu species in wines, Pohl and Sergiel found that 27–77% of the total content were polar and non-cationic forms of Cu, containing relatively strong complexes of Cu with various flavonoids and other polyphenols [144]. In fractionation and speciation studies, Karadjova et al. came to similar conclusions [57]. Based on electrochemical analysis and FAAS, Azenha and Vasconcelos concluded that Cu was present in Port wines in forms of higher molecular weight compounds with apolar character than in table wines [133]. The role of polyphenols in Cu complexation in red wines was studied by Vasconcelos et al. using ion selective electrode potentiometric titrations, leading to similar conclusions on binding of Cu with organic ligands [129]. 6.5. Zn, Sn and Mn More than 60% of Zn in wine was found to be present as positively charged labile ion and less than 15% as a complex with polyphenols, whereas negatively charged species of Zn were not found [57]. Companys et al. measured the free Zn concentration in wines using voltammetric techniques and found that the free Zn concentration represented about 5% of the total Zn concentration in the wine samples analyzed [130]. Organotin speciation in wines by solid phase microextraction and gas chromatography with flame photometric detection has been reported by Heroult et al. [145]. The applicability and suitability of the method for wine analysis was demonstrated and contamination by butyl-, phenyl- and octyltins was found in the wines analyzed [135]. In a recently published paper Sun et al. reported on the use of capillary electrophoresis coupled with ICPMS for the determination of trimethyltin, tributyltin, dibutyltin and monobutyltin in wines [146]. Furthermore, the suitability of SPE and FAAS for Mn partitioning in wines has been demonstrated by Pohl, who found that Mn is predominantly present in the form of cationic species [147]. 7. Conclusions Atomic spectrometry techniques are useful tools for elemental wine analysis in order to check toxicity, quality assurance or authenticity. Selection of a technique depends on analyte concentration level, the number of analytes to be measured and matrix effects caused by wine components. Wine matrix effects can be efficiently mitigated by the appropriate selection of the sample pretreatment method, calibration strategy, instrumentation configuration and experimental conditions. Despite the numerous applications of elemental wine analysis by means of atomic spectrometry techniques, there is still room for further improvements. First of all, new developments are required to improve sample throughput. Current sample pretreatment methods are successful to mitigate matrix effects but at the expense of sample throughput. This issue can probably be addressed by means of new devices for on-line sample pretreatment or direct wine analysis. On the other hand, new advances in instrumentation also play a major role mitigating matrix effects.

31

Thus, for instance, spectral interferences could be addressed by means of continuum source AAS or collision cell coupled to quadrupole-ICPMS. Finally, more attention should be paid to speciation studies to guarantee that the chemical form of the analyte is preserved though the whole analytical procedure. Isotopic dilution is a powerful tool to avoid non-spectral matrix effects.

References [1] http://www.oiv.int/uk/accueil/ (last visited 7 December 2010). [2] R.S. Jackson, Wine Science: Principles and Applications, 3rd ed., Elsevier, London, 2008. [3] M. Victoria Moreno-Arribas, M. Carmen Polo, Wine Chemistry and Biochemistry, 1st ed., Springer, New York, 2009. [4] X.S. Ough, M.A. Amerine, Methods for Analysis of Musts and Wines, 1st ed., Wiley & Sons, New York, 1988. [5] Compendium of International Methods of Analysis of Wine and Musts, International Organisation of Vine and Wine (OIV), Paris, 2011. [6] Methods for the analysis of wines, COMMISSION REGULATION (EEC) N◦ 2676/90 of 17 September 1990. [7] M.D. Luque de Castro, J. González-Rodríguez, P. Pérez-Juan, Food Rev. Int. 21 (2005) 231. [8] E. Yip, C.-W. Lam, W.-F. Tongm, Trends Anal. Chem. 27 (2008) 460. ˜ [9] J.G. Ibanez, A. Carreon-Alvarez, M. Barcena-Soto, N. Casillas, J. Food Comp. Anal. 21 (2008) 672. [10] P. Pohl, Trends Anal. Chem. 26 (2007) 941. [11] B. Medina, S. Augagneur, M. Barbaste, F.E. Grousset, P. Buat-Menard, Food Addit. Contam., Part A 17 (2007) 435. [12] S. Catarino, M. Madeira, F. Monteiro, F. Rocha, A.S. Curvelo-Garcia, R. Bruno de Sousa, J. Agric. Food Chem. 56 (2008) 168. [13] R. Lara, S. Cerutti, J.A. Salonia, R.A. Olsina, L.D. Martinez, Food Chem. Toxicol. 43 (2005) 293. [14] S. Galani-Nikolakaki, N. Kallithrakas-Kontos, A.A. Katsanos, Sci. Total Environ. 285 (2002) 155. [15] P.L. Teissedre, M. Krosniak, K. Portet, F. Gasc, A.L. Waterhouse, J.J. Serrano, J.C. Cabanis, G. Cros, Food Addit. Contam., Part A 15 (1998) 585. [16] F.F. Lopez, C. Cabrera, M.L. Lorenzo, M.C. Lopez, Sci. Total Environ. 220 (1998) 1. [17] M.A. Brescia, V. Caldarola, A. De Giglio, D. Benedetti, F.P. Fanizzi, A. Sacco, Anal. Chim. Acta 458 (2002) 177. [18] A. Jos, I. Moreno, A.G. González, M. López-Artíguez, A.M. Cameán, Eur. Food Res. Technol. 218 (2004) 448. [19] L. Sauvage, D. Frank, J. Stearne, M.B. Millikan, Anal. Chim. Acta 458 (2002) 223. ˜ [20] M. Castineira, R. Brandt, N. Jakubowski, A.T. Andersson, J.T. Andersson, J. Agric. Food Chem. 52 (2004) 2953. [21] D. Cozzolino, M.J. Kwiatkowski, R.G. Dambergs, W.U. Cynkar, L.J. Janik, G. Skouroumounis, M. Gishen, Talanta 74 (2008) 711. [22] A. Santalad, R. Burakham, S. Srijaranai, K. Grudpan, Microchem. J. 86 (2007) 209. [23] P.L. Buldini, S. Cavalli, J. Lal Sharma, J. Agric. Food Chem. 47 (1999) 1993. [24] O. Zerbinati, F. Balduzzi, V. Dell’Oro, J. Chromatogr. A 881 (2000) 645. ´ [25] K. Pyrzynska, Crit. Rev. Anal. Chem. 34 (2004) 69. [26] M. Aceto, O. Abollino, M.C. Bruzzoniti, E. Mentasti, C. Sarzanini, M. Melandrino, Food Addit. Contam., Part A 19 (2002) 126. [27] E. Yip, C.-W. Lam, W.-F. Tongm, Trends Anal. Chem. 28 (2009) 214. [28] C.R.C. Almeida, M.T.S.D. Vasconcelos, M. Barbaste, B. Medina, Anal. Bioanal. Chem. 374 (2002) 314. [29] C.M. Almeida, M.T.S.D. Vasconcelos, Anal. Chim. Acta 463 (2002) 165. [30] G. Grindlay, J. Mora, L. Gras, M.T.C. de Loos-Vollebregt, Anal. Chim. Acta 652 (2009) 154. [31] M. Barbaste, B. Medina, J.P. Perez-Trujillo, FAC 20 (2003) 141. ˜ [32] M.M. Castineira, R. Brandt, A. von Bohlen, N. Jakubowski, Fresenius J. Anal. Chem. 370 (2001) 553. [33] C.R. Quétel, S.M. Nelms, L. Van Nevel, I. Papadakis, D.P. Taylor, J. Anal. At. Spectrom. 16 (2001) 1091. [34] J.C.J. Da Silva, S. Cadore, J.A. Nobrega, N. Baccan, Food Addit. Contam., Part A 24 (2007) 130. [35] G. Grindlay, J. Mora, S. Maestre, L. Gras, Anal. Chim. Acta 629 (2008) 24. [36] I.M. Moreno, D. González-Weller, V. Gutierrez, M. Marino, A.M. Cameán, A.G. González, A. Hardisson, Microchem. J. 88 (2008) 56. [37] A. Gonzálvez, S. Armenta, A. Pastor, M. de la Guardia, J. Agric. Food Chem. 56 (2008) 4943. [38] A. Gonzálvez, A. Llorens, M.L. Cervera, S. Armenta, M. de la Guardia, Food Chem. 112 (2009) 26. [39] J. Kristl, M. Veber, M. Slekovec, Anal. Bioanal. Chem. 373 (2002) 200. [40] M. Kim, Food Addit. Contam., Part A 21 (2004) 154. ´ T. Stafilov, Microchim. Acta 147 (2004) [41] I. Karadjova, S. Arpadjan, J. Cvetkovic, 39. [42] M.B. Dessuy, M.G.R. Vale, A. Souza, S.L.C. Ferreria, B. Welz, D. Katskov, Talanta 74 (2008) 1321. [43] Z. Ajtony, N. Szoboszlai, E. Susko, P. Mezei, K. Gyorgy, L. Bencs, Talanta 76 (2008) 627.

32

G. Grindlay et al. / Analytica Chimica Acta 691 (2011) 18–32

[44] M.D. Huang, H. Becker-Ross, S. Florek, U. Heitmann, M. Okruss, Anal. Bioanal. Chem. 382 (2005) 382. [45] S.L.C. Ferreira, A.S. Souza, G. Brandao, H.S. Ferreira, W.N.L. dos Santos, M.F. Pimentel, M.G. Vle, Talanta 74 (2008) 699. [46] M.L. Carvalho, M.A. Barreiros, M.M. Costa, M.T. Ramos, M.I. Marques, X-Ray Spectrom. 25 (1996) 29. [47] X. Gruber, P. Kregsamer, P. Wobrauschek, C. Streli, Spectrochim. Acta, Part B 61 (2006) 1214. [48] M.J. Anjos, R.T. Lopes, E.F.O. de Jesus, S. Moreira, R.C. Barroso, C.R.F. Castro, Spectrochim. Acta, Part B 58 (2003) 2227. [49] S. Frias, C. Diaz, J.E. Conde, J.P. Pérez Trujillo, Food Addit. Contam., Part A 20 (2003) 237. [50] W. Chuachuad, J.F. Tyson, J. Anal. At. Spectrom. 20 (2005) 282. [51] L. Elci, Z. Arslan, J.F. Tyson, J. Hazard. Mater. 162 (2009) 880. [52] M. Segura, Y. Madrid, C. Cámara, J. Anal. At. Spectrom. 14 (1999) 131. [53] I. Karadjova, L. Lampugnani, A. D’Ulivo, M. Onor, D. Tsalev, Anal. Bioanal. Chem. 388 (2007) 801. [54] R.C. de Campos Costa, A.N. Araújo, Anal. Chim. Acta 438 (2001) 227. [55] M. Tuzen, M. Soylak, Anal. Chim. Acta 504 (2004) 325. [56] M. Tuzen, M. Soylak, L. Elci, Anal. Chim. Acta 548 (2005) 101. [57] I. Karadjova, B. Izgi, S. Gucer, Spectrochim. Acta, Part B 57 (2002) 581. [58] V. Lemos, M. de la Guardia, S. Ferreira, Talanta 58 (2002) 475. [59] Y. Baircioglu, S. Rio Segade, E.R. Yourd, J.F. Tyson, Anal. Chim. Acta 485 (2003) 9. [60] D. Schiavo, J. Neira, J.A. Nobrega, Talanta 76 (2008) 1113. [61] H.R. Eschnauer, L. Jakob, H. Meierer, R. Neeb, Microchim. Acta 95 (1989) 291. [62] G. Thiel, K. Danzer, Fresenius J. Anal. Chem. 357 (1997) 553. [63] C. Baluja-Santos, A. López, M.I. Basadre, A. González, Fresenius J. Anal. Chem. 355 (1996) 240. [64] G.P.G. Freschi, C.S. Dakuazu, M. de Moraes, J.A. Nobrega, J.A. Gomes Neto, Spectochim. Acta, Part B 56 (2001) 1987. [65] B.T. Kildahl, W. Lund, Fresenius J. Anal. Chem. 354 (1996) 93. [66] C. Baluja-Santos, A. González-Portal, Talanta 39 (1992) 329. [67] K. T˘asev, I. Karadjova, T. Stafilov, Microchim. Acta 149 (2005) 55. [68] M.Y. Perez-Jordan, J. Soldevila, A. Salvador, A. Pastor, M. de la Guardia, J. Anal. At. Spectrom. 13 (1998) 33. [69] F. Bianchi, M. Careri, M. Maffini, A. Mangia, C. Mucchino, Rapid Commun. Mass Spectorm. 17 (2003) 251. [70] P.P. Coetzee, F. Vanhaecke, Anal. Bioanal. Chem. 383 (2005) 977. [71] S. Augagneur, B. Medina, J. Szpunar, R. Lobinski, J. Anal. At. Spectrom. 11 (1996) 713. [72] N. Jakubowski, R. Brandt, D. Stuewer, H.R. Eschnauer, S. Görtges, Fresenius J. Anal. Chem. 364 (1999) 424. [73] M. Barbaste, L. Halicz, A. Galy, B. Medina, H. Emteborg, F.C. Adams, R. Lobinski, Talanta 54 (2001) 307. [74] X. Tian, H. Emteborg, M. Barbaste, F.C. Adams, J. Anal. At. Spectrom. 15 (2000) 829. [75] M. Barbaste, K. Robinson, S. Guilfoyle, B. Medina, R. Lobinski, J. Anal. At. Spectrom. 17 (2002) 135. [76] C.M.R. Almeida, M.T.S.D. Vasconcelos, Food Chem. 85 (2004) 7. [77] C.R.C. Almeida, M.T.S.D. Vasconcelos, J. Anal. At. Spectrom. 14 (1999) 1815. [78] C.R.C. Almeida, M.T.S.D. Vasconcelos, J. Anal. At. Spectrom. 16 (2001) 607. [79] J.L. Todolí, J.M. Mermet, Spectochim. Acta Part B 54 (1999) 895. [80] D.G. Weir, M.W. Blades, J. Anal. At. Spectrom. 9 (1994) 1311. [81] D.G. Weir, M.W. Blades, J. Anal. At. Spectrom. 9 (1994) 1323. [82] D.G. Weir, M.W. Blades, J. Anal. At. Spectrom. 11 (1996) 43. [83] D.G. Weir, M.W. Blades, J. Anal. At. Spectrom. 11 (1996) 1011. [84] K. Kahen, B.W. Acon, A. Montaser, J. Anal. At. Spectrom. 20 (2005) 631. [85] J. Goossens, T. Desmaele, L. Moens, R. Dams, Fresenius J. Anal. Chem. 347 (1993) 119. [86] M. Wasilewska, W. Goessler, M. Zischka, B. Maichin, G. Knapp, J. Anal. At. Spectrom. 17 (2002) 1121. [87] I. Rodushkin, F. Ödman, P.K. Appelblad, J. Food Comp. Anal. 12 (1999) 243. [88] S. Wangkarn, S.A. Pergantis, J. Anal. At. Spectrom. 14 (1999) 657. [89] C.R.C. Almeida, M.T.S.D. Vasconcelos, Anal. Chim. Acta 396 (1999) 45. [90] M. Aramendia, M. Resano, F. Vanhaecke, Anal. Chim. Acta 648 (2009) 23. [91] H. Wu, Y. Jin, M. Luo, S. Bi, Anal. Sci. 23 (2007) 1109. [92] I.B. Karadjova, L. Lampugnani, M. Onor, A. D’Ulivo, D.L. Tsalev, Spectochim. Acta Part B 60 (2005) 816. [93] Y. Madrid, C. Cámara, Analyst 119 (1994) 1647. [94] J.L. Capelo, H.A. Pedro, A.M. Mota, Talanta 61 (2003) 485. [95] J.L. Todolí, V. Hernandis, L. Gras, J. Mora, J. Anal. At. Spectrom. 17 (2002) 142. ˜ [96] M. Castineira, I. Feldmann, N. Jakubowski, J.T. Andersson, J. Agric. Food Chem. 52 (2004) 2962. [97] A. Jos, I. Moreno, A.G. Gonzlez, G. Repetto, A.M. Camean, Talanta 63 (2004) 377.

[98] P.P. Coetzee, F.E. Steffens, R.J. Eiselen, O.P. Augustyn, L. Balcaen, F. Vanhaecke, J. Agric. Food Chem. 53 (2005) 5060. [99] S. Catarino, A.S. Curvelo-Garcia, R. Bruno de Sousa, Talanta 70 (2006) 1073. [100] G. Thiel, G. Geisler, I. Blechschmidt, Anal. Bioanal. Chem. 378 (2004) 1630. [101] M. Olalla, M.C. González, C. Cabrea, M.C. López, J. AOAC Int. 83 (2000) 189. [102] M. Olalla, M.C. González, C. Cabrea, R. Gimenez, M.C. López, J. AOAC Int. 85 (2002) 960. [103] M. Iglesias, E. Besalua, E. Anticoa, J. Agric. Food Chem. 55 (2007) 221. [104] J. Szpunar, P. Pellerin, A. Maarov, T. Doco, P. Williams, B. Medina, R. Lobinski, J. Anal. At. Spectrom. 13 (1998) 749. [105] J. Sperova, M. Suchánek, Food Chem. 93 (2005) 659. [106] K. T˘asev, I. Karadjova, T. Stafilov, Microchim. Acta 149 (2005) 149. ´ S. Arpadjan, I. Karadjova, T. Stafilov, Spectrochim. Acta, Part B 57 [107] J. Cvetkovic, (2002) 1101. [108] I. Narin, M. Tuzen, M. Soylak, Talanta 63 (2004) 411. [109] W.N.L. dos Santos, G.C. Brandao, L.A. Portugal, J.M. David, S.L.C. Ferreira, Spectrochim. Acta, Part B 64 (2009) 601. [110] M.J. Harmse, R.I. McCrindle, J. Anal. At. Spectrom. 17 (2002) 1311. [111] J.J. Thompson, R.S. Houk, Appl. Spectrosc. 41 (1987) 801. [112] J. Goossens, L. Moens, R. Dams, Anal. Chim. Acta 293 (1994) 171. [113] P. Serapinas, P.R. Venskutonis, V. Aninkeviˇcius, Z. Eˇzerinskis, A. Galdikas, V. ˙ Food Chem. 107 (2008) 1652. Juzikiene, [114] M. Aramendía, M. Resano, F. Vanhaecke, Anal. Chim. Acta 648 (2000) 23. [115] A.A. Almeida, M.L. Bastos, M.I. Cardoso, M.A. Ferreira, J.L.F.C. Lima, M.E. Soares, J. Anal. At. Spectrom. 7 (1992) 1281. [116] http://www.titrivin.com (last visited 7 December 2010). [117] http://www.fapas.com/quality-control-materials/ (last visited 7 December 2010). [118] P. Kment, M. Mihaljevic, V. Ettler, O. Sebek, L. Strnad, L. Rohlova, Food Chem. 91 (2005) 157. ´ [119] K. Pyrzynska, Chem. Spec. Bioavailab. 19 (2007) 1. [120] A.J Mcinnon, G.R. Scollary, Talanta 44 (1997) 1649. ˜ [121] M.M. Castineira, P. Burba, N. Jakubowski, J. Andersson, Anal. Bional. Chem. 376 (2003) 174. [122] Z. Murányi, L. Papp, Microchem. J. 60 (1998) 134. [123] M.T. Pérez-Corona, M. Sánchez-Martínez, M.J. Valderrama, M.E. Rodríguez, C. Cámara, Y. Madrid, Food Chem. 124 (2011) 1050. [124] J.A. Caruso, R.G. Wuilloud, J.C. Altamirano, W.R. Harris, J. Toxicol. Environ. Health, Part B 9 (2005) 41. ˜ ˜ [125] N. Campillo, R. Penalver, P. Vinas, I. Lopez-Garcia, M. Hernandez-Cordoba, Talanta 77 (2008) 793. [126] J. Meija, S. Mounicou, J.A. Caruso, J. AOAC Int. 87 (2004) 205. [127] S.L.C. Ferreira, H.S. Ferreira, R.M. de Jesus, J.V.S. Santos, G.C. Brandao, A.S. Souza, Anal. Chim. Acta 602 (2007) 89. [128] G. Weber, Fresenius J. Anal. Chem. 340 (1991) 161. [129] M.T.S.D. Vasconcelos, M. Azenha, V. de Freitas, J. Agric. Food Chem. 47 (1999) 2791. [130] E. Companys, M. Naval-Sanchez, N. Martínez-Micaelo, J. Puy, J. Galceran, J. Agric. Food Chem. 56 (2008) 8296. [131] R. Lobinski, C. Witte, F.C. Adams, P.L. Teissedre, J.C. Cabanis, C.F. Boutron, Nature 370 (1994) 24. [132] P.L. Teissedre, R. Lobinski, M. Cabanis, J. Szpunar, J.C. Cabanis, F.C. Adams, Sci. Total Environ. 153 (1994) 247. [133] M.A.G.O. Azenha, M.T.S.D. Vasconcelos, J. Agric. Food Chem. 48 (2000) 5740. [134] M.A.G.O. Azenha, M.T.S.D. Vasconcelos, Food Chem. Toxicol. 38 (2000) 899. [135] C. Herce-Pagliai, I. Moreno, G. González, M. Repetto, A.M. Cameán, Food Addit. Cont., Part A 19 (2002) 542. [136] S. Wangkarn, S.A. Pergantis, J. Anal. At. Spectrom. 15 (2000) 627. [137] C.M. Moreira, F.A. Duarte, J. Lebherz, D. Pozebon, E.M.M. Flores, V.L. Dressler, Food Chem. 126 (2011) 1406. [138] A. Bakar Tawali, G. Schwedt, Fresenius J. Anal. Chem. 357 (1997) 50. [139] E.K. Paleologos, D.L. Giokas, S.M. Tzouwara-Karayanni, M.I. Karayannis, Anal. Chim. Acta 458 (2002) 241. ˇ [140] R. Ajlec, J. Stupar, Analyst 114 (1989) 137. ´ T. Stafilov, Food Control 17 [141] K. Tasev, I. Karadjova, S. Arpadjan, J. Cvetkovic, (2006) 484. [142] P. Pohl, B. Prusisz, Talanta 77 (2009) 1732. ˜ J.A. Barrero-Moreno, M.C. Perez-Conde, Talanta 51 (2000) [143] P. Pulido-Tofino, 537. [144] P. Pohl, I. Sergiel, J. Agric. Food Chem. 57 (2009) 9378. [145] J. Heroult, M. Bueno, M. Potin-Gautier, G. Lespes, J. Chrom., A 1180 (2008) 122. [146] J. Sun, B. He, Y. Yin, L. Li, G. Jiang, Anal. Methods 2 (2010) 2025. [147] P. Pohl, Food Chem. 114 (2009) 996.