Multi-element analysis of South African wines and their provenance soils. by ICP-MS and their classification according to geographical origin using

Transcription

1 Multi-element analysis of South African wines and their provenance soils by ICP-MS and their classification according to geographical origin using multivariate statistics. by Gert van der Linde DISSERTATION Submitted in fulfillment of the requirements for the degree MAGISTER SCIENTIAE in CHEMISTRY in the FACULTY OF SCIENCE at the UNIVERSITY OF JOHANNESBURG SUPERVISOR: Prof P. P. Coetzee CO-SUPERVISOR: Dr. J.L. Fischer NOVEMBER 2008

2

3 Acknowledgements. I would like to thank: Prof Paul Coetzee for his time, effort and endless patience. NRF for financial support Dr. G.J. van der Linde, Margaret van der Linde and Marlene van der Linde for their help at each stage of this project and checking the spelling. Dr. Herman Niekerk for all of his help and technical expertise. Prof F. Steffens for the statistical calculations of the final results. Nick Downer for his assistance during analysis and the dark times. Amanda Austin for her support and faith during the years.

4 Table of Contents. ABSTRACT LIST OF ABBREVIATIONS LIST OF TABLES LIST OF FIGURES i ii iv v CHAPTER 1: INTRODUCTION 1 CHAPTER 2: LITERATURE SURVEY Different analyses for point of origin determination Sr isotopic ratios Pb analysis B isotopic ratios Rare Earth Elements (REE) Multi-Element analysis Instrumentation Atomic Absorption Spectroscopy (AAS) Nuclear Magnetic Resonance (NMR) Inductively Coupled Plasma Optical Emission 10 Spectroscopy/ Atomic Emission Spectroscopy (ICP-OES). 2.3 Inductively Coupled Plasma Mass Spectrometry (ICP- 11 MS) Sample preparation Solid samples Liquid samples Nebulisers Spray Chamber 16

5 2.3.4 Torch and plasma characteristics Plasma Interface Collision Cell Technology (CCT) and Dynamic Reaction 20 Cells (DRC) Collision Cell Technology (CCT) Dynamic Reaction Cells (DRC) Quadrupole-based ICP-MS Detectors Internal standards Statistical analysis Visual inspection Principle Component Analysis (PCA) Stepwise Discriminant Analysis. 27 CHAPTER 3: METHOD DEVELOPMENT Method and materials Reagents.and standards Instrumentation Samples and sample preparation Method validation Wine dilution method Microwave digestion method for wine Grape juice dilution Grape juice microwave method Soil sample investigation Isoptopic abundance test. 75

6 3.3 Conclusions. 75 CHAPTER Classification of wines and soils Classification of wines Classification of soils Classification of a, b and c soil samples Correlation between wine and soil. 92 CHAPTER 5: CONCLUSION AND FUTURE WORK 98 REFERENCES 100 APPENDIX A: Additional method development graphs 110 APPENDIX B: Statistical calculations of PCA and Stepwise 137 discriminant analysis

7 ABSTRACT The South African wine industry is well respected internationally for producing high quality wines. The possible adulteration of these wines can lead to loss of reputation and a loss of sales and could also be dangerous to consumer s health. Multi-element analysis of wines is one way of implementing quality control and the same multi-element data can also be used to prove the point of origin. The metal content of the fruit (grapes) should represent the metal content of the soil in which the plants (vineyards) were grown. An Inductively Coupled Plasma Mass Spectrometer (ICP-MS) was used with correct internal standard and interference correction to obtain reliable concentrations for 27 elements (Li, B, Al, Sc, V, Cr, Mn, Co, Ni, Cu, Zn, Se, Rb, Sr, Zr, Nb, Mo, Cd, Sn, Sb, Ba, Ce, Nd, W, Tl, Pb and U) of 1:1 diluted wines and microwave digested vineyard soil from four South-African wine-producing regions: Stellenbosch, Swartland, Robertson and Walker Bay. This multi-element data was then interpreted using multivariate statistical analysis in order to determine which elements have the ability to discriminate between the four regions. Li, B, Sc, Ni, Mn, Co, Cu, and Rb were the elements that were identified to have discrimination ability. 96% of wines and 100% of vineyard soils were correctly classified. Indirectly it has been proven that the metal content of the soil can be correlated to the metal content of the wine. This methodology can be reliably used in industry for quality control and routine provenance determination Keywords: Dilution, Inductively Coupled Plasma (ICP-MS), internal standard, microwave digestion, multivariate statistical analysis, principle component analysis (PCA), stepwise discriminant analysis, wine. i

8 LIST OF ABBREVIATIONS REE: Rare Earth Elements AAS: Atomic Absorption Spectrometry FAAS: Flame Atomic Absorption Spectrometry ETAAS: Electro-Thermal Atomic Absorption Spectrometry NMR: Nuclear Magnetic Resonance SNIF-NMR: Site-specific Natural Isotope Fractionisation ICP-OES: Inductively Coupled Plasma Optical Emission Spectrometry ICP-MS: Inductively Coupled Plasma Mass Spectrometry LA: Laser Ablation TDS: Total Dissolved Solids CCT: Collision Cell Technology DRC: Dynamic Reaction Cell m/z Mass to charge ratio ST: Stellenbosch ii

9 SW: Swartland R: Robertson W: Walker-Bay PCA Principle Component Analysis iii

10 LIST OF TABLES. Table 1: Wines from four regions selected for this project. 32 Table 2: Microwave digestion program. 34 Table 3: Grapes from the Stellenbosch region. 35 Table 4: Results of percentage TDS of microwave soil extraction. 62 Table 5: Average element concentrations (µg/l) and standard deviations in red and white wines from Stellenbosch, Swartland, Robertson and Walker Bay regions. 80 Table 6: Average element concentrations (µg/l) and standard deviations in provenant soils from Stellenbosch, Swartland, Robertson, and Walker Bay regions. 81 iv

11 LIST OF FIGURES. Figure 1: Plasma Interface. Copied from Montaser, Figure 2: Quadrupole mass filter. Copied from Montaser, Figure 3: Recoveries for the analysis of ST1 L Avenir 2X dilution using different internal standards 40 Figure 4: Recoveries for the analysis of ST2 Delheim 2X dilution using different internal standards. 41 Figure 5: Recoveries for the analysis of ST12 Simonsig 2X dilution using different internal standards. 42 Figure 6: Recoveries for the analysis of ST1 L Avenir 150µL HNO 3 microwave digestion using different internal standards. 46 Figure 7: Recoveries for the analysis of ST2 Delheim 150µL HNO 3 microwave digestion using different internal standards. 47 Figure 8: Recoveries for the analysis of ST12 Simonsig 150µL HNO 3 microwave digestion using different internal standards. 48 Figure 9: Recoveries for the analysis of ST1 L Avenir 600µL HNO 3 microwave digestion using different internal standards 50 Figure 10: Recoveries for the analysis of ST2 Delheim 600µL HNO 3 microwave digestion using different internal standards 51 Figure 11: Recoveries for the analysis of ST12 Simonsig 600µL HNO 3 52 v

12 microwave digestion using different internal standards Figure 12: Recoveries for the analysis of Simonsig grape juice 2X dilution using different internal standards 54 Figure 13: Recoveries for the analysis of Delheim grape juice 2X dilution using different internal standards 55 Figure 14: Recoveries for the analysis of Simonsig grape juice150µl HNO 3 microwave digestion using different internal standards 57 Figure 15: Recoveries for the analysis of Delheim grape juice 150µL HNO 3 microwave digestion using different internal standards 58 Figure 16: Recoveries for the analysis of Simonsig grape juice 600µL HNO 3 microwave digestion using different internal standards 60 Figure 17: Recoveries for the analysis of Delheim grape juice 600µL HNO 3 microwave digestion using different internal standards 61 Figure 18: Recoveries for the analysis of ST13 Groot Constantia 10X dilution using different internal standards 64 Figure 19: Recoveries for the analysis of ST17 Alexanderfontein 10X dilution using different internal standards 65 Figure 20: Recoveries for the analysis of SW17 Porterville 10X dilution using different internal standards 66 Figure 21: Recoveries for the analysis of SW19 Porterville 10X dilution using different internal standards 67 vi

13 Figure 22: Recoveries for the analysis of R8 Clairvaux 10X dilution using different internal standards 68 Figure 23: Recoveries for the analysis of R16 Bonnievale 10X dilution using different internal standards 69 Figure 24: Recoveries for the analysis of W1 Beaumont 10X dilution using different internal standards 70 Figure 25: Canonical Discriminant Functions plot of wines from the four vineyards. 84 Figure 26: 3D Canonical Discriminant Functions plot of wines from the four vineyards. 85 Figure 27: Discriminant functions of Red wines and White wines from the four vineyards. 86 Figure 28: Canonical Discriminant Functions plot of soils from the four vineyards. 87 Figure 29: 3D Canonical Discriminant Functions plot of soils from the four vineyards. 88 Figure 30: Discriminant functions of Red wine soils and White wine soils from the four vineyards. 89 Figure 31: Canonical Discriminant Functions plot of a soils from the four vineyards. 90 vii

14 Figure 32: Canonical Discriminant Functions plot of b soils from the four vineyards 91 Figure 33: Canonical Discriminant Functions plot of c soils from the four vineyards 92 Figure 34: Canonical Discriminant Functions plot of wines with specifically selected elements from the four vineyards. 94 Figure 35: 3D Canonical Discriminant Functions plot of wines with specifically selected elements from the four vineyards. 95 Figure 36: Canonical Discriminant Functions plot of soils with specifically selected elements from the four vineyards. 96 Figure 37: 3D Canonical Discriminant Functions plot of wines with specifically selected elements from the four vineyards. 97 viii

15 CHAPTER 1 INTRODUCTION Wine: The alcoholic beverage obtained from the fermentation of the juice of freshly gathered grapes, the fermentation taking place in the district of origin according to local tradition and practice (Robinson, 1994). This introduction chapter strives to show the importance of wine and other foodstuff analysis and authentication as well as stating the objectives of this project. South Africa produces internationally known wines and commands respect from its overseas competitors. The 2005 wine output was about million litres which makes South Africa the 9 th largest wine producer in the world (that is about 3% of the world s wine). In 2006 the South African wine industry earned $ million in sales from the production of wine. Add to this total the sales of grapes for wholesalers, wine for brandy, distilling wine and grape juice and the total increases to $ million (SAWIS, 2008). All these figures show that agricultural products are of great economical importance. The quality of these products are also of concern in order to protect the consumer (Christaki and Tzia, 2002). One of the quality factors that must be kept in check is the multi-element content of the foodstuffs. High concentrations of certain elements have severe toxic effects on the human body and these metals for example Cd, As, Pb and Hg, should be monitored at all times. This list will increase with time as environmental contamination becomes of greater concern. This continued analysis of foodstuffs (including wine) may also indicate a source of contamination near the areas of production and processing which includes aerosol and industrial pollution (Schwartz and Heking, 1991; Stroh et al., 1994; Miller-Ihli, 1996; Thiel and Danzer, 1997; Greenough et al., 1997; Pérez-Jordán et al., 1998; Castiñeira et al., 2001; Nikolakaki et al., 2002; Almeida et al., 2002; Sauvage et al., 2002; Barbaste et al., 2002; Taylor et al., 2003; Almeida et al., 1

16 2003 Gómez et al., 2004; Thiel et al., 2004; Gómez et al., 2004; Coetzee et al., 2005). Analysis of foodstuff may not only be used for quality control but can be employed for certifying point of origin. In theory the metal content of the plant and fruit should reflect that of the soil in which they are grown (Stroh et al., 1994; Almeida and Vasconcelos, 2003; Moreda-Piñeiro et al., 2003; Thiel et al., 2004; Gómez et al., 2004; Coetzee et al., 2005). Proving the origin of agricultural products and wines in particular, is of great importance to the agricultural industry. The vineyards of South Africa are well known internationally and adulteration of these wines can lead to a loss of prestige and sales (Robinson, 1994). Previous studies have shown that provenance determination of wines(schwartz and Heking, 1991; Stroh et al., 1994; Miller-Ihli, 1996; Thiel and Danzer, 1997; Greenough et al., 1997; Pérez-Jordán et al., 1998; Castiñeira et al. 2001; Nikolakaki et al., 2002; Almeida et al., 2002; Sauvage et al., 2002; Barbaste et al., 2002; Taylor et al., 2003; Almeida et al., 2003; Gómez et al., 2004a; Thiel et al., 2004; Gómez et al., 2004b; Coetzee et al., 2005), fruit juices (Ogrinc et al., 2003), olive oil (Ogrinc et al., 2003), and tea (Moreda-Piñeiro et al., 2003) can be accomplished. There are a variety of instruments that can be used to determine the multielement content of wines such as Atomic Absorption (AA) instruments (Miller-Ihli, 1996; Lara et al., 2005) and Inductively Coupled Plasma Optical Emission Spectrometer (ICP-OES) (Thiel and Danzer, 1997; Lara et al., 2005). The ICP-OES s counterpart is the Inductively Coupled Plasma Mass Spectrometer (ICP-MS). This mass spectrometer has a low Limit of Detection (LOD) and is also multi-element capable. Sample throughput is also very high and once the mass interferences are understood and compensated for, this instrument delivers reliable data (Evans and Giglo, 1993; Montaser, 1998; Tangen and Lund, 1999). The ICP-MS is the best instrument for analysis of Rare 2

17 Earth Elements (REE) (Augagneur et al., 1996; Jakubowski et al., 1999) and other elements with low concentrations that are of great importance for provenance determination (Nikolakaki et al., 2002; Gómez et al., 2004; Coetzee et al., 2005; Taylor et al., 2003; Thiel et al., 2004; Sauvage et al., 2002; Castiñeira et al., 2001; Gómez et al., 2004; Almeida et al., 2002; Almeida and Vasconcelos, 2003). An ICP-MS was used for this project to analyse the wine and vineyard soils. Objectives of this project were: Method development for reliable routine multi-element analysis of South- African wines and soils (from four different regions). Using multivariate statistical analysis (Principle Component Analysis (PCA) and step wise discriminant analysis) to interpret wine and soil multielement results. Identify elements that have the ability to discriminate between the wines and soils from different geographical origin. Investigate the behaviour of different internal standards with different sample matrices. 3

18 CHAPTER 2 LITERATURE SURVEY This chapter summarises what has been done by other researchers in the field and stresses the importance of quality control of foodstuffs, food authentication, and provenance determination with the emphasis on wine. Methodologies and instrumentation to accomplish provenance determination is also discussed. All wines are produced using the same basic process that has changed little over the last few centuries (Robinson, 1994; Ribéreau-Gayon et al., 2000). Wine production has a number of steps some of which are possible sources of contamination (Christaki and Tzia, 2002). Metal surfaces in contact with the juice and wine during the vinification process could, for example, constitute a source of metal contamination. Other sources of contamination are already present in the vineyard in the form of agri-chemicals. Contamination may therefore originate from the environment, equipment in the production process, and the workers involved in the production (Robinson, 1994; Rosman et al., 1998; Rodushkin et al., 1999; Almeida and Vasconcelos, 1999; Ribéreau-Gayon et al., 2000; Barbaste et al., 2001; Frías et al., 2001 Christaki and Tzia, 2002; Almeida and Vasconcelos, 2003a; Almeida and Vasconcelos, 2003b; Gómez et al., 2004; Coetzee et al., 2005; Kment et al., 2005; Minnaar and Booyse, 2006). To ensure that the consumer is protected and possible sources of contamination are traced, quality control checks have to be brought into the production cycle (Robinson, 1994; Arvanitoyannis et al., 1999; Ribéreau-Gayon et al., 2000; Christaki and Tzia, 2002; Almeida and Vasconcelos, 2003a; Almeida and Vasconcelos, 2003;). This is true not only for wine but for many other foodstuffs, beverages and recreational products (Schwartz and Heking, 1991; Miller-Ihli, 1996; Watling, 1998; Rodushkin et al., 1999; Hernández-Caraballo et al., 2003; Moreda-Piñeiro et al., 2003). 4

19 Another reason to analyse wine and other foodstuffs, is to prove their point of origin. Suitable analyses can also indicate if the wine has been adulterated. Some of the practices of adulterating or doctoring are legal when it is necessary such as the addition of sugar to wines before fermentation to increase total alcohol in the wine (Martin, 1990; Robinson, 1994; Ribéreau-Gayon et al., 2000; Ogrinc et al., 2003). These techniques are, however, illegal when employed to deliberately mislead the customer. The adulteration of wine and other foodstuffs may even be dangerous to the customer. Examples of this were investigated by Rosman et al., 1998 and include the use of lead in ancient times, methanol contamination in the 20 th century, and the addition of antifreeze (ethylene glycol), as described by The Oxford Companion to Wine (Robinson, 1994; Christaki and Tzia, 2002; Sauvage et al., 2002; Hernández-Caraballo et al., 2003; Ogrinc et al., 2003). Ogrinc et al., 2003 also indicated the same is also true with fruit juices and olive oil. Certain trace metal elements are necessary for growth of all plants and their fruit (Robinson, 1994; Ribéreau-Gayon et al., 2000; Tagliavini and Rombolà, 2001; van der Linde, 2008). The only place where these elements could originate is the soil from which plants obtain their nourishment. As stated by Coetzee et al., 2005 and Almeida et al., 2003 the elemental content found in wine and agricultural products should in principle reflect the geochemistry and thus the point of origin of the agricultural product (Kwan et al., 1979; Moret et al., 1994; Augagneur et al., 1996; Watling, 1998; Almeida and Vasconcelos, 1999; Jakubowski et al., 1999; Moreda-Piñeiro et al., 2003; Gómez et al., 2004; Kment et al., 2005). A few problems do arise with this type of reasoning in determining the point of origin of wines and other foodstuffs. As mentioned above there are many sources of contamination to be found during all of the steps of production. According to the Oxford Companion to Wine (Robinson, 1994) and the Handbook of Enology (Ribéreau-Gayon et al., 2000), wine and grape must (crushed grapes) 5

20 are somewhat acidic and will leach metals from any winemaking equipment made from metal alloys (e.g. bronze in pipes and taps, stainless steel fermentation vats). The same was also observed by Rodushkin et al., 1999 and Hernández-Caraballo et al., 2003 with the manufacture of other alcoholic beverages, especially illegal moonshine. The time that grape must is in contact with the crushed grape skins and seeds will also affect the metal content. Grape skins and seeds contain a higher concentration of trace elements than the juice. The use of fertilisers and pesticides is necessary to ensure a high crop yield but they are also sources of metal contamination (Robinson, 1994; Minnaar and Booyse, 1996; Rosman et al., 1998; Almeida and Vasconcelos, 1999; Rodushkin et al., 1999; Ribéreau-Gayon et al., 2000; Frías et al., 2001; Castiñeira et al., 2001; Barbaste et al., 2001; Christaki and Tzia, 2002; Almeida and Vasconcelos, 2003a; Almeida and Vasconcelos, 2003b; Gómez et al., 2004; Kment et al., 2005; Coetzee et al., 2005). Some wines, in particular everyday white wines are filtered using bentonite fining (filtering with clay for clarification and removal of proteins in wines and musts before or after fermentation) (Robinson, 1994). Jakubowski et al., 1999 indicated bentonite fining will remove most of the unwanted components and enhance the flavour and body of wines, but metals will definitely leach out of the clay into the wines (Robinson, 1994; Ribéreau- Gayon et al., 2000). These are the most common of obstacles that are encountered when attempting to prove the point of origin of many agricultural products. 2.1 Different analyses for point of origin determination There are different types of analysis that can be conducted to prove the point of origin of wines and other foodstuffs Sr isotopic ratios Almeida and Vasconcelos, 2001; 2004 and Barbaste et al., 2002 have conducted extensive research on the use of Sr isotopic ratios for point of origin 6

21 determinations. They have proven that even with some influences from the vinification process, the Sr 87 /Sr 86 isotopic ratio is reliable for point of origin determinations. There are some problems that need to be overcome before this technique can be effectively applied. Rb 87 is an isobaric mass interference with Sr 87. All analysts overcame this problem by separating the two isotopes by ion exchange. It has proven to work but the more a sample is handled by the analyst the greater the risk of sample contamination becomes. The sample preparation also becomes much more laborious and expensive. These disadvantages aside, this method has been proven to work and can be used to complement other methods of analysis (Almeida and Vasconcelos, 2001; (Almeida and Vasconcelos 2004; Barbaste et al., 2002) Pb isotope ratios Pb is very useful in the determination of age and geographical location of geological samples but could be problematic for wine and foodstuffs (Almeida and Vasconcelos, 1999). Pb is a contaminant in most cases and originates from a wide array of sources. Rosman et al., 1998, Barbaste et al., 2001 and Almeida and Vasconcelos, 1999; 2003 showed that the possible sources of Pb are emissions of vehicles, brass in the pipes of the winery and Pb introduced by older vinification methods are some of. That is why for this project Pb was left out of the final statistical calculations to ensure that the final point of origin calculation could not be influenced by contamination. Analysis of Pb is still useful for quality control of the final wine in industry (Rodushkin et al., 1999) B isotopic ratios Another isotope ratio that is popular to use is B 11 /B 10. Some of the recent work on this isotope was conducted by Coetzee and Vanhaecke, 2005 to investigate the differences between South-African, Italian and French wines. Coetzee and Vanhaecke have proven that this isotope ratio could be successfully used in differentiating between wines from different areas. Some of the disadvantages are that the sample has to be diluted considerably in order to reduce sample 7

22 matrix interferences. In order to observe these slight differences the instrument must have a very high sensitivity and precision. Sample throughput is low due to long analysis time. Further research is being done to use this method in order to apply it much more readily to wine analysis Rare Earth Elements (REE) Jakubowski et al., 1999 has shown that REE could be readily detected in wine and can be used for point of origin determinations. Very similar to multi-element analysis but in this case the main focus is a specific set of elements. These REE s have similar chemical characteristics and can also be influenced by the vinification process just like other elements. The interference of the sample matrix and the very low concentrations make the detection and use of these elements also very difficult. With further advances in instrumentation this disadvantage could be overcome and REE s could become part of normal analysis of wines (Augagneur et al., 1996) Multi-element analysis Greenough et al., 1997; Frias et al., 2001; Barbaste et al., 2002; Almeida and Vasconcelos, 2003; Thiel et al., 2004; Coetzee et al., 2005; Minnaar and Booyse, 2006 have proven that multi-element analysis of wines for point of origin is the easiest and most reliable method available. Instead of relying on only a selected few elements the analyst now has a wide range to choose from. This means that elements that are susceptible to contamination and interferences can be ignored. Multi-element analysis requires less sample preparation than those previously mentioned and sample through-put is very high. Multi-element analysis can also be done with a variety of instruments, meaning that it can be accomplished at a lower cost. If a method of analysis is to be successful in industry then it must be fast, cost-effective and reliable. The work in this project strived to duplicate Coetzee et al., 2005 work and build further on it by analysing more wines and also the soil from the vineyards.in order to prove the link between the wine and its provenant soil. The other mentioned analysis (2.1.1 Sr isotopic ratios;

23 Pb analysis; B isotopic ratios; Rare Earth Elements (REE)) can be used to supplement the multi-element data to ensure that the final answer can be compared with different methods (Scarponi et al., 1982; Baluja-Santos and Gonzalez-Portal, 1992; Stroh et al., 1994; Miller-Ihli, 1996; Baxter et al., 1997; Thiel and Danzer, 1997; Pérez-Jordán et al., 1998; Frías et al., 2001; Castiñeira et al., 2001; Sauvage et al., 2002; Almeida and Vasconcelos, 2002; Almeida et al., 2002; Nikolakaki et al., 2002; Taylor et al., 2003; Taylor et al., 2003; Gómez et al., 2004; Kment et al., 2005; Lara et al., 2005; Minnaar and Booyse, 2006). 2.2 Instrumentation The type of instrument will influence the type of elements to be investigated and the ability to detect different concentrations of these elements. In the following sections (2.2.1 Atomic Absorption Spectroscopy, Nuclear Magnetic Resonance, Inductively Coupled Plasma Optical Emission Spectroscopy) different instruments that were used in other projects will be discussed while the instrumental principles of the Inductively Coupled Plasma Mass Spectrometer (ICP-MS) are discussed in Section 2.3 Inductively Coupled Plasma Mass Spectrometry (ICP-MS) Atomic Absorption Spectroscopy (AAS) Atomic spectroscopy is based on using a flame or electro thermal (graphite furnace) atomiser for atomisation. In flame atomisation (FAAS), the sample, in aqueous solution is nebulised and transported into the flame by the gas flow. Measurement is achieved by emission or absorption spectra. For electro thermal atomisation (ETAAS) a graphite furnace is employed to atomise the sample and absorption or emission spectra can then be obtained. The detection limit for FAAS is in the mg/l range and the ETAAS is from 0.1µg/L to 10µg/L. These instruments are very robust and not prone to much interference and the interferences that are there can be easily corrected. With hydride generation AAS, a hydride is formed from which absorption spectra can be obtained. Baluja-Santos and Gonzalez-Portal, 1992 investigated the application of hydride 9

24 generation for wine and other beverages. Its detection limit is also in the µg/l range but it is only effective for one element at a time. The detection limit of FAAS is not low enough to observe the metals useful for the determination of point of origin due to possible metal contamination during production. However, it can work very well for quality control of the production cycle (Moret et al., 1994; Skoog et al., 1996; Rebolo et al., 2000; Frías et al., 2001; Sauvage et al., 2002; Nikolakaki et al., 2002; Lara et al., 2005) Nuclear Magnetic Resonance (NMR) NMR is used to detect and identify a number of low molecular mass units in complex mixtures. The sample preparation is also very simple and nondestructive. This technique as shown by Martin, 1990 and Košir and Kidric, 2002 can also detect adulteration by organic materials, such as sugar in chaptalisation (i.e. the practice of adding sugar to the juice prior to fermentation to increase the potential alcohol and quality of the wine). To observe this adulteration using other methods for the detection of metals is impossible. The recording of deuterium ( 2 H) spectra at their natural abundance level can be achieved using a new method called Site-specific Natural Isotope Fractionation studied by NMR (SNIF-NMR). There is a significant difference in the distribution of deuterium in a molecule such as ethanol. Using SNIF-NMR it is possible to determine the point of origin. This method may be combined with other analytical techniques such as ICP-MS to obtain a complete answer on geographical origin and any possible adulteration (Martin et al., 1988; Day et al., 1994; Ebbing, 1996; Ogrinc et al., 2003) Inductively Coupled Plasma Optical Emission Spectroscopy/ Atomic Emission Spectroscopy (ICP-OES) The use of argon (Ar) plasma as the atomisation source began in the mid 1970s and has several advantages over flame atomisers. For one the temperature of the plasma is K. This is much higher than any flame atomiser can ever hope to achieve. The end result is that all of the analyte is atomised and excited 10

25 and the emission spectra can then be obtained. ICP-OES instruments are very robust, reliable and are multi-element instruments with very low detection limits (bordering on the sub µg/l level depending on the element). Minnaar and Booyse, 2006, Thiel and Danzer, 1997 and Lara et al., 2005 have shown that the ICP-OES can be successfully applied to multi-element analysis of wines. Sample preparation is easy, the obtained spectra are relatively simple depending on the wavelength being observed and there are also few interferences. Boorn and Browner, 1982 did show that organic solvents like the type encountered in wines can influence the plasma but can be compensated for with careful modification of instrument parameters (Kwan et al., 1979; Schwartz and Heking, 1991; Miller-Ihli, 1996; Skoog et al., 1996). 2.3 Inductively Coupled Plasma Mass Spectrometry (ICP-MS) A standard ICP-MS instrument uses the same sample introduction system (peristaltic pump, pneumatic nebuliser and spray chamber) and ICP torch (atomisation and excitation source) as an ICP-OES instrument. However that is where the similarities end. Instead of observing and focusing emitted light from the plasma in an ICP-MS it is the ions that are focused by electric fields and a detector counts the individual ions. The result is a mass spectrum that has to be interpreted. These instruments have a very broad detection range (from ng/l to low mg/l) making them the most powerful available for trace and ultra trace analysis. It is a multi-element method with the ability to obtain isotopic data as well (Moret et al., 1994; Rosman et al., 1998; Rodushkin et al., 1999; Barbaste et al., 2001; Almeida and Vasconcelos, 2001; Barbaste et al., 2002; Almeida and Vasconcelos, 2003; Almeida and Vasconcelos, 2004). Alternative sample introduction systems and chromatographic separation techniques can also be combined with an ICP-MS with relative ease. Greenough et al., 1997, Barbaste et al., 2002, Almeida and Vasconcelos, 2003, Thiel et al., 2004, Coetzee et al., 2005 have proven time and again the usefulness of ICP-MS in determining the point of origin and quality control of wine and many other foodstuffs (Martin et al., 1988; Stroh et al., 1994; Day et al., 1994; Goossens et al., 1994; Augagneur et 11

26 al., 1996; Baxter et al., 1997; Rosman et al., 1998; Watling, 1998; Pérez-Jordán et al., 1998; Almeida and Vasconcelos, 1999; Rodushkin et al., 1999; Vanhaecke et al., 1999; Jakubowski et al., 1999; Castiñeira et al., 2001; Barbaste et al., 2001; Almeida and Vasconcelos, 2001; Barbaste et al., 2002; Almeida and Vasconcelos, 2002; Almeida et al., 2002; Almeida and Vasconcelos, 2003; Moreda-Piñeiro et al., 2003; Taylor et al., 2003; Marengo and Aceto, 2003; Niemelä et al., 2003; Gómez et al., 2004; Gómez et al., 2004; Almeida and Vasconcelos, 2004; Coetzee and Vanhaecke, 2005; Kment et al., 2005) ICP-MS has proven its worth when analysing metals, but what about negative ions such as the halogens? Fulford and Quan, 1988 and Vickers et al., 1988 proved it to be possible but investigations by Niu and Houk, 1996, have shown only average sensitivity with this method. Cl and F are also very hard to convert efficiently to positive ions. At the normal sampling position all elements will be present in their ionic form (M + ). This means that the amount of negative ions would not be as abundant in an ICP when compared with other ions (Chtaib and Schmit, 1988; Montaser, 1998; Pupyshev and Surikov, 2004). This instrument, however, has many drawbacks that should be considered. Instrument cost is one major disadvantage. An ICP-OES costs between $ to $ while an ICP-MS could well be over $ for a low resolution instrument and more than $1 million for a high resolution instrument at the time of writing (Oki, 2008). The operating costs should also be brought into the equation and an ICP-MS has many more components and thus parts that can fail (such as the torches, the sampler/ skimmer cones, vacuum pump, detectors and vacuum seals). Because of their sensitivity to the environment these instruments must be located in the correct laboratory. Ambient temperature should not change more than 2 C per 30 min (Oki, 2008). Ideally ICP-MS s should operate in a clean lab environment. An analyst must follow stringent sample preparation methods to ensure that no contamination interferes with the analysis. The tuning of an ICP- MS is also complicated and requires an experienced operator to obtain a stable 12

27 signal. ICP-MS is very susceptible to matrix effects which mean that sample preparation is very important. Mathematical correction can be added to compensate for this during analysis. However the interferences in ICP-MS are quite extensive and it is sometimes impossible to completely compensate for some interferences (Montaser, 1998; Evans and Giglo, 1993; van Niekerk, 2008; Downer, 2008) Sample preparation Sample preparation plays a very important part in the successful analysis of trace metal content for safety purposes and point of origin determination. The type of preparation used depends on the sample and the instrument. Most instruments require a homogeneous solution for analysis (FAAS, ETA, ICP-OES/MS, NMR). This is true for all environmental, biological and geological samples. The first and easiest way is to dilute the sample with solution used for analysis which usually is 1% HNO 3. This method is useful for its ease of use, speed and small chance for contamination. The less the sample is handled the less chance for any contamination or loss of analyte. A dilution factor of 1:1 was preferred by Coetzee et al., This ensures that most of the elements are well within detection limits. The disadvantage is that the matrix interferences are much higher and correcting measures have to be applied to ensure accurate analysis, for example: matrix matched calibration standards and standard addition (Boorn and Browner, 1982; Stroh et al., 1994; Augagneur et al., 1996; Greenough et al., 1997; Baxter et al., 1997; Pérez-Jordán et al., 1998; Rosman et al., 1998; Jakubowski et al., 1999; Rodushkin et al., 1999; Castiñeira et al., 2001; Sauvage et al., 2002; Almeida and Vasconcelos, 2002; Almeida and Vasconcelos, 2003; Almeida and Vasconcelos, 2003; Moreda-Piñeiro et al., 2003; Taylor et al., 2003; Almeida and Vasconcelos, 2004; Thiel et al., 2004; Minnaar and Booyse, 2006). In this project the performance and practicality of microwave digestion and dilution of wine samples were compared. While both methods do work the results of the dilution method was much better. Microwave digestion is 13

28 susceptible to contamination and the amount of sample that is needed would make some elements fall below detection limit. Soil samples can only prepared by microwave and other solid sample preparation methods which will be described in section Solid samples Solid samples An older method available is the open beaker hot-plate dissolution. This method has been used for many decades and is the easiest and cheapest method available to an analyst for dissolving solid samples. The sample is weighed in a vessel and acid added and then heated on a hot plate. Disadvantages of this method are long digestion times (hours to days), loss of analyte and a greater chance of contamination (Energlyn and Brealey, 1971). One other method that may be applied to liquid and solid samples is microwave digestion. A small amount of sample is weighed off in a Teflon container (designed for a microwave digestion instrument), the desired acids are added and then microwaved for the required time. This appears simple but there are some factors that have to be considered. The amount of sample required for analysis must be calculated to ensure that analyte levels are above the detection limit. The intended use of the measurements i.e. quality control, semiquantitative or quantitative analysis will determine the limits of error. The sample matrix will determine the amount and type of acid and reagents that will be used. For example, geological samples will require hydrofluoric acid (HF) to completely dissolve the silicates, or if a partial digestion is required then the HF may be omitted. Carbon-based matrices (plant material, wine and grape juice) can be easily digested using HNO 3 and H 2 O 2. Perchloric acid (HClO 4 ) is also a useful acid but must be handled with care as hot HClO 4 is unstable and could explode when heated. It is sometimes necessary to dilute this digested liquid further to ensure the acid concentration is low enough to prevent damage to instruments and to keep background signals to a minimum since everything that goes into a 14

29 plasma is ionised and may interfere with analysis as proven by Evans and Giglo, 1993 and Tanner et al., 2002 (Montaser, 1998). One of the newer methods to analyse solid samples is the use of laser ablation (LA). It is essentially a high powered-laser, with optics for beam control and steering with an interface connection to carry the ablated or vaporised material into the ICP-MS/OES (Montaser, 1998). This method has been successful on a wide range of solid samples from geological, biological, and archaeological samples. Normally a relatively large sample size is necessary but with LA only a small sample area is ablated. With art it is absolutely necessary to minimise the damage as some artworks are centuries old and are worth a fortune, unless it s the falsified one. Law enforcement agencies are now turning to these types of analysis for forensic investigations (Watling, 1998; Jackson et al., 2003; Boulyga and Heumann, 2005; Smith et al., 2005) Liquid samples Nebulisers Matching the correct nebuliser with the sample type is very important. For wine analysis the best would be the concentric nebuliser. That said the wine must be diluted to ensure that the nebuliser is not clogged and that a stable signal can be obtained. For partial digested soil samples a cross-flow or concentric nebuliser would function perfectly but once again the Total Dissolved Solids (TDS) is very high and every precaution must be taken to ensure no blockages occur (Montaser, 1998). Liquid sample introduction is one of the most common and user friendly methods available. The liquid sample must first be turned into an aerosol by a nebuliser before it can be transported into an ICP. There is quite a large selection available and each one has its own advantages and disadvantages. Only the most commonly used nebulisers will be discussed here. 15

30 The first is the cross-flow nebulisers. The gas flow is directed at a right angle at the tip of a capillary tube to the flow of the sample liquid. The V-groove nebuliser is also considered a cross-flow nebuliser and can even function with slurries. These nebulisers are much more effective at preventing blockages when sample liquids have very high TDS and are far more rugged for routine work than most nebulisers. These nebulisers are not very efficient at producing small droplets for use in ICP-MS analysis (Skoog et al., 1996; Montaser, 1998). Concentric nebulisers are better suited for ICP-MS analysis. The sample liquid flows in a capillary tube parallel with the gas flow towards a low pressure region created by the gas flow. The droplet size is much smaller and more consistent in comparison to cross-flow nebulisers. They also provide very good sensitivity and stability provided that samples have a low TDS content. To ensure that the nebuliser does not get blocked it is advisable to filter samples (Skoog et al., 1996; Montaser, 1998). One of the latest nebulisers available is the micro flow nebuliser. They work on the same principle as the concentric nebuliser but at higher gas pressure and lower sample flow rate. When limited sample is available and memory effects are problematic then these nebulisers are ideal. These nebulisers get blocked even easier than the normal concentric nebuliser (Skoog et al., 1996; Montaser, 1998) Spray Chamber Connected to the nebuliser is the spray chamber which ensures that only droplets of minimum size are allowed to be transported into the plasma. If droplets are bigger than 10µm then efficient desolvation, volatilisation, and atomisation will not be possible in the plasma. This will result in signal instability and unwanted background noise. Spray chambers also smooth any pulses generated by the peristaltic pump. For mass spectrometry work spray chambers are cooled to ensure that the minimum amount of sample is introduced into the 16

31 plasma. This also results in reduction of oxide levels and allows an analyst to aspirate volatile organic solvents. Adding an impact bead in the spray chamber will break up more of the droplets which increase the number of droplets allowed to enter the plasma which enhances signal intensity (Montaser, 1998). To ensure signal stability then the appropriate spray chamber must be matched with the type of sample that is being analysed. With wine samples a Scott-type or impact bead type would be the best. They work very well with samples that are organic or have high TDS content. A chilled spray chamber was also used for this project. The Scott-type or double-pass spray chamber is one of the most used and rugged spray chambers available. Droplets are removed by turbulent deposition and gravity and are then subsequently removed by a tube for drainage. The fine droplets are forced back into the central part of the chamber by the positive pressure in the drain and are then transported into the plasma (Montaser, 1998). Another spray chamber is the cyclonic type. With this spray chamber the larger droplets are removed by centrifugal forces and deposited on the walls of the chamber. The smaller droplets are then transported into the plasma. This type of spray chamber is not commonly used but they have a greater sampling efficiency for clean samples. When organic and high TDS samples are used then this efficiency will decline (Montaser, 1998). The impact bead type spray chamber is much simpler than any of the previous. A glass bead is placed in the path of the aerosol from the nebuliser. The droplets are then broken up more. The impact bead may also be added to any of the other spray chambers to increase their efficiency Torch and plasma characteristics. The basic design of the ICP torch has changed little and the ones used for OES instruments and those of MS are very similar. A conventional Fassel torch is 17

32 used for Ar ICP-MS instruments. The torch consists out of three concentric quartz tubes. Through these tubes the Ar flows but each has its own use. The gas in the outer tube is for cooling, containing the plasma and preventing the melting of the torch. Gas flowing through the intermediate tube is required for the stability and formation of the plasma. In the central tube the injector gas flow transports the sample aerosol into the plasma (Montaser, 1998). Plasmas are very dense systems in which the gasses are accelerating and expanding outward, usually perpendicular to the gas flow. To insert the sample aerosol into this system is very difficult. That is why the annular and doughnut base shape created by the outer gas flow in an ICP discharge is so important. This shape allows the insertion of the sample aerosol into the centre of the plasma to ensure efficient and complete desolvation, vaporisation, atomisation, excitation, and ionisation. It is thus important to have the tip of the injector tube as close to the base of the plasma as possible. This will ensure that the maximum amount of sample aerosol is forced into the plasma. The plasma itself consists out of three zones: the initial radiation zone, normal analytical zone and the plasma tail. In the initial radiation zone, analyte is converted into mostly molecular species and neutral atoms. In the normal analytical zone the analyte is converted into atomic anlyte ions. The plasma tail contains neutral atoms and oxides formed by reactions of ions and oxygen from the air (Niu and Houk, 1996; Montaser, 1998). Plasma characteristics will change depending on the type of solvent aspirated. Most samples are aqueous and usually acidic. Plasma conditions are well known when these types of matrices are used. Some samples like diluted wine have ethanol in them. Boorn and Browner, 1982 showed that plasma conditions do change but not much and these changes in the plasma can be compensated for. If there is too much organic solvents in the plasma then carbon can be deposited on the sampler cone. Sometimes this does not really interfere but adding some oxygen to the plasma gas will enhance the conversion of the C to CO 2. These 18

33 problems must be considered for any type of analysis especially diluted beverages (Boorn and Browner, 1982; Vickers et al., 1988; Evans and Giglo, 1993; Montaser, 1998) Plasma Interface The interface is the region where the conditions of the plasma that are at K temperatures and normal pressure is connected to the lenses and detector which is at room temperature and high vacuum. Figure 1 is a simplified representation of the plasma interface. Figure 1: Plasma Interface. Copied from Montaser, The interface region has always presented problems to the entire principle of ICP-MS. The interface consists of two funnel like components called cones. These cones are usually manufactured out of Pt or Ni. The first cone is called the sampler cone and has an orifice of around 1mm in diameter. If the sample has a high TDS content then a salt build-up can be observed around the orifice. Prolonged use of cones with such a salt build-up could lead to memory effects and reduced sensitivity. The region between the cones is known as the expansion chamber. Here the plasma expands supersonically which results in large distances between the different particles. Due to the large distances there 19

34 is no longer any interaction between the particles and in effect freezing the ion population as it existed in the plasma (Montaser, 1998). Some problems arise in the interface zone. The first one is known as secondary radio frequency (RF) discharge. An electrostatic coupling is a characteristic of all ICP s but this is a very serious problem for ICP-MS. The electrostatic discharge is between the plasma and the sampler cone. The result of this undesired discharge is a high continuum background, the cone s lifetime is reduced and unwanted orifice ions are also created. The orifice ions consist of the metal from which the cones are manufactured. To compensate for this the induction coils can be physically grounded. Another problem in the interface region is the space charge effect. Plasmas are neutral by nature but charge diffusion and electrostatic effects cause charge separation and lead to the charge flow becoming predominantly positive. This net charge imbalance exerts an electric field which can lead to defocusing of the ion beam. Bias against low-mass elements, a bias towards high mass elements in isotopic ratio analysis, suppression of analyte signal by higher mass concomitant elements, severe suppression by higher mass concomitant elements and more suppression of lower ion signals by concomitant elements are some of the distortions caused by the space charge effect. In order to minimise the space charge effect the ion current is reduced while retaining substantial neutrality or to retain charge neutrality for as far as possible into the ion optics. There are still further studies being done on this subject but space charge effects will always hamper ICP-MS analysis no matter what sample matrix is used (Evans and Giglo, 1993; Niu and Houk, 1996; Montaser, 1998) Collision Cell Technology (CCT) and Dynamic Reaction Cells (DRC) Some elements are considered problematic when using ICP-MS for their analysis. These elements suffer greatly from spectral interferences derived from ions created in the plasma itself. For example 35 Cl 16 O interferes with 51 V; 40 Ar 12 C 20

35 interferes with 52 Cr, and the most notorious one 40 Ar 16 O interferes with 56 Fe. There are many more but these are the best known ones. In order to reduce these interferences there are some steps that can be taken. The type of sample matrix can be changed. Or sample preparation methodology may also be changed, for example: HCl not used for dissolution or ensure that the matrix does not contain any carbon. Another method is to use cool plasma conditions which produce fewer of these Ar-based interferences. These methods do work but they are cumbersome and time consuming (Montaser, 1998; Tanner et al., 2002). Using a CCT/ DRC improves detection limits and serves as an extra focusing lens for the ion beam. Most instruments only have a CCT or only a DRC built in. This means an analyst must carefully plan the sample preparation method to ensure that the interferences can be dealt with by the cell. The Thermo Electron X-Series 2 used for this project has both of these cells combined into one. This is the best possible combination which allows the analyst to use the interference reducing powers of both. Wine samples also have a wide variety of elements and components in them which have the potential of forming interfering ions. It is thus an advantage to use a combined CCT and DRC for wine and beverage analysis (Evans and Giglo, 1993; Montaser, 1998; Niemelä et al., 2003; Tanner et al., 2002) Collision Cell Technology (CCT) Ions are extracted as normal but they are focussed into a multipole (usually hexapole) collision cell array positioned before the quadrupole mass analyser and pressurised with a collision cell gas. This creates a controllable atmosphere for gas phase reactions and collisions. Inside this cell there are collisions between the different ions and molecular ions produced by the plasma and the collision gas which results in removal of nearly all argon-based molecular ion interferences. The one major problem encountered with this system is that secondary collisions do take place and these unwanted products must be eliminated. Using a system of kinetic energy discrimination for this purpose has 21

36 proven very useful. The collision cell bias is slightly less positive than the mass filter bias. The collision cell product ions have the same energy as the cell bias is rejected. The analyte ions have higher energy than the cell and are allowed through to the mass filter (quadrupole) Dynamic Reaction Cells (DRC) The dynamic reaction cells use a different operating principle. A reaction cell is placed in the path of the ion beam just before the quadrupole mass filter and pressurised with a reaction cell gas. Exactly like the collision cell just described but that is where all similarities end. More energetic reaction gasses such as NH 3 and CH 4 can now be used to ensure that ion molecule reactions take place. The interfering ions react with the gas and transform them into neutral species or other species that are different from the analyte ions. Secondary unwanted reactions that produce unwanted interferences may also take place within the cell. This can be avoided by careful manipulation of the electric field within the cell to eliminate this problem. This is advantageous when unknown sample types are introduced into the plasma because the instrument settings can be adjusted in real time to compensate for any unwanted reactions and interferences (Montaser, 1998; Oki, 2008) Quadrupole-based ICP-MS There are many types of mass filter equipped ICP-MS systems available. Some of the most used ones are the low resolution quadrupole ICP-MS, double focusing sector field high resolution ICP-MS, and the time of flight ICP-MS. The Thermo Electron X-series 2 ICP-MS used in this project is a quadrupole-based instrument and this type will be further discussed here. Ions are extracted from the plasma, focussed and transported to the quadrupole mass filter in order to obtain the desired isotope for analysis. A quadrupole consists of four round rods that are at equal distance from each other. 22

37 Figure 2: Quadrupole mass filter. Copied from Montaser, Figure 2 gives a simplified representation of a quadrupole and the direction that extracted ions follow. A voltage is applied to the rods with a positive to the rods in the y axis and negative to the rods in the x axis. If this voltage is kept stationary then the ions will collide with one of the rods and be lost. That is why the potential is reduced and then reversed. This changing of the potential will result in spinning the ions down the length of the rods into the detector. The ions will be accelerated with a given applied potential. If the ion is accelerated to a speed that is too fast or slow to steer onto a stable flight path during the potential change, then the ion will collide with one of the quadrupole rods and be lost. Only ions with the required mass to charge ratio (m/z) will make it through the quadrupole to the detector to be counted. This required m/z ratio can be obtained by subtle variations in the applied potential of the quadrupole rods. This entire system is kept at a very high vacuum to ensure that ion beam is stable and that unwanted collisions with anything else but the rods or detector be avoided 23