摘要:

Critical Review Phenolic compounds in olives Danielle Ryan and Kevin Robards* Charles Sturt University Riverina, PO Box 588, Wagga Wagga 2678, Australia Summary of contents Introduction Structure of plant phenolics or more hydroxy substituents Role of phenolics in olives Properties and function Phenolics as antioxidants Phenolics and fruit quality Factors affecting the phenolic profile of olives Varietal influences Other factors Olive development and maturation Processing and storage Oil production Table olives Analysis Sample preparation Quantification Chromatographic methods Liquid chromatography Detection Keywords: Phenolic compounds; olives; mass spectrometry; chromatography; fruit Introduction Archaeological evidence traces olive trees back to 6000bc and frequent Biblical references appear throughout the New Testament.The Mediterranean region nowadays serves as the major international olive growing area, accounting for almost 98% of the world’s olive tree plantation.1 Olives are rarely consumed as a natural fruit due to their extreme bitterness, but rather are consumed in either one of two forms; as oil or table olives.The significance of the olive oil market in the olive industry is apparent as it consumes approximately 90% of the annual production of olives. Annual world consumption of olive oil in 1995/96 was 1777 thousand tonnes and this increased from 1508 thousand tonnes in 1970/71 but most striking has been the rapid growth in consumption2,3 in high-income countries like Japan, the United States, Canada and Australia.The increasing health consciousness of today’s more cosmopolitan society explains the rising consumption of olive oil around the world and hence the rapid growth of the olive industry. The beneficial health properties of olive oil have been known for centuries, particularly in the Mediterranean region. Olives and olive oil are an inherent part of Mediterranean culture and diet, and hence the decreased incidence of cardiovascular disease in this area (being one of the lowest in the Western Hemisphere) has been attributed to their consumption. 4 These effects have been attributed5 to the high content of oleic acid in olive oil, which serves to slow down the penetration of fatty acids into the arterial walls.The preventative superiority of olive oil is also attributed to its antioxidant composition; namely, tocopherols and phenolic compounds.The latter are highly diverse,6–8 both in their chemical structure and proposed biological functions. Their metabolic pathways are particularly complex with, in many cases, multiple alternative metabolic fates. Profiling of the components of a pathway over time provides a dynamic view of the metabolic events occurring in the plant. As stated by Amiot et al.:9 ‘Knowledge of the variations in phenolic compounds should make it possible first to obtain better understanding of the relationships that may exist between these substances and the physiology and organoleptic qualities of the fruit and second to provide a more solid basis for processing techniques, thus leading to improved quality.’ The intense interest in plant phenolics which has been manifested over several decades, accounts for the many reviews and monographs10–12 devoted to various aspects of these compounds.For example, the role of plant phenolics in the prevention and treatment of disease has been examined.13 Food sources of plant phenolics have been reviewed14 and the same authors discussed the biological activities and functions of phenolic compounds, especially as they relate to their mechanisms of anticarcinogenicity.This review critically examines the analytical chemistry of the phenolics in olives. Methods used for the analysis of samples other than olives will be discussed where these illustrate current applications which can be extended to include olives or which may emerge as important advances over existing methods.The review also addresses the factors that impact upon the phenolic composition of olive fruits and oils. Structure of plant phenolics The plant phenols are aromatic secondary metabolites that embrace a considerable range of substances possessing an Kevin Robards is Associate Professor of chemistry at Charles Sturt University Riverina. He obtained his PhD. in analytical chemistry form the University of New South Wales in 1979.His research interests are focused on the application of analytical chemistry to food science and in particular the identification and role of naturally occurring phenolic compounds in fruits. Analyst, May 1998, Vol. 123 (31R–44R) 31Raromatic ring bearing one or more hydroxy substituents.15 In the present context, this definition is not entirely satisfactory since it inevitably includes compounds such as oestrone, the female sex hormone (which is principally terpenoid in origin).For this reason, a definition based on metabolic origin is preferable, the plant phenols being regarded as those substances derived from the shikimate pathway and phenylpropanoid metabolism (Fig. 1). Phenolic compounds present in olives are conventionally characterised as ‘polyphenols’, an unfortunate term since not all are polyhydroxy derivatives. In particular, a number of compounds, namely, cinnamic acid, elenolic acid, shikimic acid and quinic acid, are treated in the present discussion as phenolics because of metabolic considerations although they lack a phenolic group or even an aromatic ring.Plant phenols have been classified15 into 15 major groupings distinguished by the number of constitutive carbon atoms in conjunction with the structure of the basic phenolic skeleton. The range of known phenolics is thus vast but of the various groups only the benzoic acids, cinnamic acids, flavonoids and iridoids (Table 1) are of major significance in olives.Additional structural complexity is introduced by the common occurrence of certain phenolics as the O-glycosides in which one or more of the phenolic hydroxy groups is bound to a sugar or sugars by an acid-labile hemiacetal bond. Glucose is the most commonly encountered sugar with rhamnose and the disaccharide, rutinose (6-O-a-l-rhamnosyl-d-glucose) also encountered. Acylation of the glycosides in which one or more of the sugar hydroxys is derivatised with an acid, such as acetic or ferulic acid, is occasionally observed.Phenolic compounds associated with olives are listed in Table 2 with some representative structures shown in Fig. 2. Role of phenolics in olives Metabolic pathways involving phenolics are complex with, in many cases, multiple alternative metabolic fates for a given metabolite which may vary markedly from tissue to tissue, from one growth condition to another, and in response to environmental stimuli.Hence, establishing a biological function for such compounds is often very difficult. Nevertheless, almost all of the phenolic compounds possess several common biological and chemical properties; namely, antioxidant activity, the ability to scavenge both active oxygen species and electrophiles, the ability to inhibit nitrosation and to chelate metal ions, the potential for autoxidation, and the capability to modulate certain cellular enzyme activities. Properties and function In some cases, phenolic function may well be related to primary metabolism.Some phenolics have an effect on olive plant growth while others protect the more vulnerable cell constituents against photooxidation by UV light15 by virtue of their strong ultraviolet absorption.15 Hence, phenolics play a key role in fruit preservation. In general, however, the search for a function for these compounds has focused on the interaction that may take place between the plant and other living organisms and phenolics in olives are now recognized for their antimicrobial activity47, molluscicidal properties34, their preventative role in Dacus oleae infestations32 and resistance to other parasite invasions.15 Their role in disease resistance is well established.Muller48 defined phytoalexins as ‘compounds produced after infection under the influence of two metabolic systems, that of the host and that of the parasite, and inhibitory to the parasite’.Most literature reports are concerned with the role of phytoalexins in resistance to disease caused by fungi whilst fewer relate phytoalexin accumulation to resistance to disease caused by bacteria. However, Chowdhury et al.49 have reported the minimum inhibitory concentration of several simple and complex phenolics found in olive fruits against four pathogenic bacteria. Caffeic acid was the most effective agent although oleuropein, the major phenolic constituent of olives also exhibited50 bactericidal action.By the mid-1960s, it became apparent that phytoalexins were produced not only in response to infection but also in response to various forms of physiological stress. A common response of plant cells to stress, such as wounding, infection or elicitation, is the induced incorporation of phenylpropanoids into the cell wall. However, as stated by Matern and Grimmig:51 ‘The precise role of the polyphenol cell wall reinforcement for the protection of plants has .. . remained ill-defined due to limited analytical knowledge and the complexity of the cell wall architecture.’ The discovery of new analytical methods underpins scientific progress and nowhere is this more evident than in the study of the phenolics. Phenolics as antioxidants As a consequence of their fundamental chemical properties, the phenolics inhibit lipid peroxidation52 and exhibit various physiological activities.53 The antioxidant properties of the phenolics are well known54–57 and continue to attract considerable research effort.Thus, plants such as the herb rosemary are highly acclaimed for their antioxidant properties,58 which have largely been attributed to the phenolic compounds carnosol, rosmanol and rosmadial.54 Similarly, the phenolics in olives have attracted attention as antioxidants.9,33 Total hydrophilic phenols and the oleosidic forms of 3,4-dihydroxyphenylethanol (hydroxytyrosol) were correlated (r = 0.97) with the oxidative stability of virgin olive oil59 whereas tocopherols showed low correlation (r = 0.05).More specifically, antioxidant activity in refined olive oil decreased60 in the series hydroxytyrosol, Fig. 1 Metabolic pathways leading to the formation of phenolic compounds. Table 1 Major classes of fruit phenolic compounds in olives Number of C Basic atoms skeleton Class Example 7 C6-C1 Benzoic acids p-Hydroxybenzoic acid Vanillic acid Protocatechuic acid 9 C6- C3 Hydroxycinnamic acids Caffeic acid 15 C6-C3-C6 Flavonoids Anthocyanins Cyanidin Flavonoid glycosides Rutin — Iridoids Oleuropein Ligstroside n Lignins Tannins 32R Analyst, May 1998, Vol. 123caffeic acid > butylated hydroxytoluene (BHT) > protocatechuic acid, syringic acid.Tyrosol, p-hydroxyphenylacetic acid, o-coumaric acid, p-coumaric acid, p-hydroxybenzoic acid and vanillic acid had very little or no antioxidant activity, and their contribution to the stability of the oil was negligible. A variety of methods are used to assess the antioxidant activity of crude olive extracts and purified phenolics.One approach involves measurement of the inhibition of oxidative deterioration of an oil or model substance, such as methyl linoleate.33 This is conveniently performed in the Rancimat apparatus,59 which has been used to demonstrate61 that the activity of tyrosol (in refined tallow) was lower than that of the synthetic BHT whereas oleuropein showed a stronger activity although the best protective effect was obtained with gallic acid esters and hydroxytyrosol.Care must be exercised in the interpretation of data relating to antioxidant activity as the substrate62 and also the analytical technique influences the results. The effect of substrate can be attributed62 to the strong influence of the unsaturation type and degree of the lipid system on the kinetics and mechanism of the antioxidative action of the phenols. For example, when tested in another accelerated oventest on refined sunflower oil thin films,61 the activity of hydroxytyrosol was lower than that of gallic acid esters.Similarly, the trends in antioxidant activity of phenolics differed63 according to whether hydroperoxide formation (peroxide value) or decomposition (hexanal and volatiles) was measured in accelerated stability tests on olive oil. These results emphasise the need to measure at least two oxidation parameters to better evaluate antioxidants and the oxidative stability of olive oils.Alternative techniques for measuring antioxidant activity include the electrochemical measurement of oxygen consumption and electron spin resonance (ESR) spin trapping.64 The latter involves generation of hydroxyl radicals by the Fenton reaction which are then trapped by 5,5-dimethyl- 1-pyrroline-Noxide in competition with the test sample. The electrochemical technique involves measurement of the oxygen depletion rate in a heterogeneous lipid/water emulsion with lipid oxidation initiated by metmyoglobin.The data relate to the effect of antioxidant on the propagation of oxidation while the ESR free radical method relates to the effect of antioxidant on the initiation step. The two techniques have not been applied to Table 2 Literature survey of phenolic compounds found in olives Molecular Phenolic compound Leaves Seed Pulp Oil mass Apigenin 16 16 (absent) 270 Apigenin-7-glycosides 26, 32 26, 44 Caffeic acid 26 (18), 26 27, 28, 37, 38, 40, 41 180 Chlorogenic acid 21 343 Cinnamic acid 27, 28, 29, 37, 41 148 Cornoside 39 316 o-Coumaric acid 17 40, 41 164 p-Coumaric acid 26 17, 18, 26, 35 20, 27, 28, 29, 31, 32, 37, 40, 41 164 Cyanidin-3-glycosides 26 26, 44 Demethyloleuropein 26 24, 26, 43 526 Elenolic acid 26 26 30 242 Elenolic acid glucoside (35), 43 36 Ferulic acid 17 20, 28, 37, 41 194 Gallic acid 17 36 170 (Halleridone) 39 154 Hesperidin 26 26 610 Homovanillic acid 38 182 p-Hydroxybenzoic acid 39, 41 138 p-Hydroxyphenylacetic acid 22 38, 39, 40 152 (p-Hydroxyphenyl)ethanol 30, 41 138 Hydroxytyrosol 7, 26 18, 26 28, 30, 31, 32, 36, 38, 41, 42, 45 154 Ligstroside 6, 7, 39, 46 34 524 Luteolin 16, 21 16 286 Luteolin-7-glucoside 21, 26, 32 9, 19, 26, 35, 44 448 Luteolin-7-rutinoside 32 Nuezhenide 23 Nuezhenide oleoside 23 Oleuropein 6, 7, 16, 26, 32, 46 9, 16, 18, 19, 25, 26, 33, 34, 35, 39, 43 28, 30, 41 540 Oleoside and oleuroside 6, 7 Protocatechuic acid 40, 41 154 Quercitin 16 16 302 Quercitin-3-rutinoside (rutin) 21 9, 35, 44 610 Salidroside 23, 39 Sinapic acid 28, 29 224 Syringic acid 17 20, 27, 28, 29, 40, 41, 45 198 Tyrosol 17, 39 27, 28, 29, 31, 36, 38, 42, 45 138 Tyrosol glucoside 39 300 Vanillic acid 17, 18 20, 27, 28, 29, 31, 32, 36, 41 168 Veratric acid 22 Verbascoside 9, 18, 19, 35 624 Analyst, May 1998, Vol. 123 33RO COOCH3 O HO HO O-Gluc O oleuropein O COOCH3 O HO O-Gluc O ligstroside O COOCH3 O O O Gluc O OH R nuezhide O C O HO HO CH3 O deacetoxyoleuropein aglycone O H O O O-CO CH2OH O HO HO HO CH CH OH OH O COOH O HO O-Gluc O demethyloleuropein HO verbascoside O COOH3 O O OH elenolic acid HO OH tyrosol OH HO HO hydroxytyrosol Gluc Gluc olives but they were compared for measuring the antioxidant activity of various spices.In olive oil, the phenolic content serves as an important qualitative parameter due to its correlation with the peroxide number, free fatty acidity, and sensorial quality.30,41 Free fatty acids (FFA) provide an index of the degree of lipase activity and when present at high concentrations, produce undesirable aromas in the oil.65 Because phenolics function as antioxidant constituents of olive oil, a high FFA content invariably indicates a high degree of lipase activity and hence a reduced antioxidant content.Similarly, peroxide number, or peroxide value (PV) monitors the initial products of oxidation; that is, the hydroperoxides.The PV therefore offers one of the most direct measures of lipid peroxidation.66 The amount of peroxides that must be formed to produce noticeable rancidity is dependent upon the composition of the oil and, in particular, the degree of unsaturation and the presence of antioxidants, notably, the phenolics. Phenolics and fruit quality Phenolic compounds may contribute to fruit quality in a number of ways; for example, by contributing to sensory attributes, such as colour and flavour, and through the contribution of some specific phenolics, in particular oleuropein,9 to the intense bitterness of the olive fruit.Other bitter phenolics occurring in the fruit include the glucosides, salidroside, nuezhenide, and nuezhenide oleoside, together with two secoiridoid glucosides of uncertain structure containing tyrosol, elenolic acid and glucose moieties, which have been identified23 in the seeds of Olea europaea. Cimato et al.42 attribute the organoleptic value and the preservability of olive oil to the phenolics and tocopherols although the phenolic compounds may also contribute to flavour in a negative sense.Thus, the ethyl ester of cinnamic acid and 4-vinylphenol was identified67 by gas chromatography–mass spectrometry (GC–MS) in the steam distillate from unacceptable olive oils. The source of the 4-vinylphenol was attributed to decarboxylation of p-coumaric acid. Phenolics can also contribute to fruit quality via their role in browning reactions.Thus, oxidation products of oleuropein, in conjunction with those of other native phenolics are known to be responsible for the characteristic black colour of mature olive fruits.1,35 Enzymatic oxidation of endogenous o-diphenols into o-quinones, which can then polymerise into brown products, results in the discolouration and softening of olive fruits, and the ultimate destruction of the product’s commercial value. The reaction is catalysed by polyphenol oxidases, which comprise a large group of enzymes all of which are characterised68 by their ability to utilise molecular oxygen during the oxidation of phenolic substrates.The browning reactions are mediated by various metal ions such as iron(iii)69 and manganese.70 The susceptibility of olives to browning can be examined using model solutions71 and illustrates the complex interactions between polyphenol oxidase activity and phenolic content. In some instances, the rate of browning has been positively correlated with both the content of oleuropein, the major substrate for the reaction, and the polyphenol oxidase activity.72 In most instances, however, the rate of reaction has been substrate limited18,73,74 with no correlation to the enzyme activity.Factors affecting the phenolic profile of olives Phenolics are characteristic constituents of green plants occurring in virtually all parts of the plant but with quantitative distributions that vary between different organs of the plant and within different populations of the same plant species.Phenolics in olive pulp and oil constitute a complex mixture, the complete chemical nature of which has not, as yet, been elucidated.66 For example, there are many phenolics present in Fig. 2 Chemical structure of representative phenolic compounds found in olives. 34R Analyst, May 1998, Vol. 123low concentrations which remain unidentified but whose significance may far outweigh their concentration level.Isolation and structure elucidation of these compounds are the initial steps to understanding their significance and action. Information on their biosynthesis is essential to understand the interaction between plants and the environment. Methods of characterisation and identification follow those in general use for natural substances. Hence, preparation of an extract, biological screening, bioguided fractionation, isolation and structure elucidation is the usual approach.For the latter, physical methods based on spectral characteristics feature prominently although older chemical and biochemical approaches should be considered particularly as adjuncts to spectral analysis. Factors contributing to the variability in phenolic distribution include the cultivar and genetics, maturity, climate, position on the tree, rootstock and agricultural practices. In the case of processed products, technological processes to which olive fruits are exposed may also impact significantly on the phenolic content.Varietal influences The olive fruit is characterised by the epicarp (skin), with a soft, pulpy flesh (mesocarp), and the endocarp (stone), which contains the seed or kernel. In ripe olives, the seed makes up some 2–3% of the total mass, the stone 13–23% and the flesh or mesocarp some 84–90% but occasionally as low as 65%. The composition of the flesh, stone and seed components is given in Table 375 but clearly the components of the flesh are quantitatively the more important.There are approximately 2500 known varieties of olives, 250 of which are classified as commercial cultivars by the International Olive Oil Council (IOOC). These commercial cultivars are used for the production of either olive oil or table olives. The particular use of a given cultivar is determined by its oil content and size, with larger fruits ( > 4 g) being favoured for table olive consumption.Olive varieties with an oil content of less than 12% such as Ascolano, Calamata and Manzanillo, are almost exclusively used for table olive production.75 Similarly, olive varieties with a high oil content are exploited for the purposes of olive oil production. The flesh components pass either as is or transformed, to the oil, which is mainly composed of triacylglycerols with small quantities of free fatty acids, glycerols, phosphatides, pigments, carbohydrates, proteins, flavour compounds, phenols, sterols and unidentified resinous substances (Table 4).The amount of these constituents varies17,27,77–79 with cultivar and environmental conditions. For example, the effects of cultivar, growth locality and extraction technology on the phenolic content (tyrosol, vanillic acid, caffeic acid, syringic acid, p-coumaric acid, and cinnamic acid) of Sardinian olive oils has been examined by multivariate statistics.27 The variety Bosana is particularly rich in phenols and within this variety there was evidence that the extraction technology exerted a considerable effect on phenolic composition.Oil components from the seed, though a minor component, still become part of the olive oil but do not have the same composition80 as that from the flesh. Other factors Although few studies have examined the effects of agronomic and climatic factors on fresh fruit, the effects particularly of rainfall and growing temperature on the characteristics of olive oil have been extensively reported.27,77,78,81 For instance, oil produced from orchards at 800 m was of better quality than the oil from an altitude of 100 m.17 The enhanced quality of oil obtained from higher elevation was largely explained by the oil’s higher oxidative stability compared to that from lower altitudes.This can be attributed to the higher tocopherol and total phenolic content of fruit harvested from the higher altitude. In another study of climatic effects, the compounds mainly affected were aliphatic alcohols, phenolics and headspace constituents,79 which are of particular importance in the sensory characterisation and quality of olive oil. Tous and Romera78 concluded that the oxidative stability of olive oil varies according to cultivar and location.Variation in phenolic content with harvesting period and its subsequent effect upon oil quality has been investigated.42 During the first harvesting period, phenolic levels were higher irrespective of environment and cultivar and gradually declined as the olives ripened.Deidda et al.81 and Alessandri et al.82 have identified a relationship between early harvesting and the production of high quality olive oil. Similarly, Garcia and coworkers 70,83 studied the oil obtained from different olive varieties, and found that most of the oils exhibited an increase in titratable acidity, responsible for the production of rancid flavours, and a decrease in total phenols as ripening progressed.The selection of an optimal harvesting date should therefore be ascertained to preserve the organoleptic properties of the oil and to prevent the production of inferior quality olive oil due to delayed harvesting. Olive development and maturation The flowering of the olive tree marks the beginning of fruit development. In the following six–eight months the olive attains its maximum fruit weight.75 This is followed by fruit colour change and associated physiological modifications, with the appearance of the purplish-black olive fruit indicating the end of olive morphology.Two degrees of maturation are recognised84 in olive fruits, namely green maturation and black maturation. Amiot et al.43 have included a third phase in olive development, aptly named the growth phase, which occurs prior to that of green and black maturation, during which the accumulation of oleuropein occurs. In contrast, four stages of maturation have been identified by Garcia and co-workers,70,83 which quite simply correspond to the apparent changes in fruit Table 3 Olive fruit composition75 Constituent Flesh Stone Seed Water 50–60 9.3 30.0 Oil 15–30 0.7 27.3 N matter 2–5 3.4 10.2 Sugars 3–75 41.0 26.6 Cellulose 3– 6 38.0 1.9 Ash 1–2 4.1 1.5 Phenolics 2–2.5 0.1 0.5–1.0 Intermediate 3.4 2.4 Table 4 Minor components (ppm) of virgin and refined olive oil (data from ref. 76, p. 30) Virgin Refined Component olive oils olive oils Phenolics and related substances 350 80 Hydrocarbons 2000 120 Squalene 1500 150 b-Carotene 300 120 Tocopherols 150 100 Esters 100 30 Aldehydes and ketones 40 10 Fatty alcohols 200 100 Terpene alcohols 3500 2500 Sterol alcohols 2500 1500 Analyst, May 1998, Vol. 123 35Rcolour and anthocyanin content at the green, spotted, purple and black stage of olive maturation. One of the difficulties associated with maturation studies is the precise identification of the various physiological stages.Some authors43 in recognition of this difficulty have plotted harvest date versus change in phenolic content. This approach makes no allowance for the vastly different rates of maturation of fruit on the same tree unless data for a given harvest date are selected also for the stage of development by fruit colour, for example. Fruit development can fortunately be characterised15 using one or more of the following criteria: (1) the appearance of new compounds, (2) the disappearance of certain compounds, (3) the occurrence of various characteristic ratios between certain compounds, and/or (4) the evolution of the activity of numerous enzymes leading to the biosynthesis or degradation of phenolic compounds. For example, green maturation is characterised by a reduction in chlorophyll content in conjunction with fruit softening and an increase in oil content. A further reduction in chlorophyll content is apparent in the black maturation phase, along with a significant increase in CO2 accumulation, ethylene secretion and anthocyanin content.Anthocyanins are responsible for black fruit colouring, and are classified as phenolics. The notion of phenolic compounds serving as biological markers of the physiological stages of growth and fruit maturation is well known.15 Cimato42 showed that with fruit ripening, hydrolysis of components with ‘higher molecular weight’ occurred, with the formation of tyrosol and hydroxytyrosol.Thus, the concentration of tyrosol and hydroxytyrosol was also shown to increase with the harvesting period, which has been correlated with an evident reduction in four unidentified, but presumably phenolic components. The majority of research on the relationship between phenolics and olive development concerns oleuropein, which is known to be the most prominent and significant individual phenolic component of olive pulp, reaching concentrations of up to 14% on a dry weight basis in young Picholine olives.9 The concentration of oleuropein declines with fruit maturity in accordance with the second criteria used to characterise fruit maturation; that is, the decrease in concentration of certain compounds.Amiot et al.43 have also shown that oleuropein degradation in olives is accompanied by the accumulation of two compounds, namely demethyloleuropein and elenolic acid glycoside, of which, only the former is phenolic.The fact that neither of these compounds was present prior to green maturation, in conjunction with the difficulty associated with identifying this stage, enabled the characterisation of the green maturation phase of olive development. It should also be noted that out of eleven cultivars examined, demethyloleuropein was only present in two cultivars.43 The idea of demethyloleuropein serving as a varietal marker has therefore been suggested. Similarly, Vlahov44 has suggested the exploitation of distinct flavonoid compositions of olive cultivars as a tool for biochemical characterisation of fruit varieties. Amiot et al.,9 have established an inverse relationship between oleuropein content in olive fruit and other phenols, such as certain flavonoids and verbascoside (a heterosidic ester of caffeic acid and hydroxytyrosol).Research conducted by Vlahov44 supports the findings of Amiot et al.,9 who observed increases in given flavonoid compounds in three olive fruit varieties with the onset of maturation.It is interesting to note that part of the verbascoside and oleuropein molecule are the same. It could therefore be hypothesised that partial degradation of the oleuropein molecule is responsible for the formation of verbascoside, since verbascoside cannot be detected in very young fruits.9 This is in agreement with the hypothesis proposed by Amiot et al.9 that ‘the successive evolution of oleuropein and verbascoside and their biochemical relationship may suggest the existence of a metabolic relationship between these two compounds’.Olive trees are known to be alternate bearers,85 providing high fruit yields one year and low yields the next. Alternate bearing causes a major problem in the olive industry, particularly in warmer climates,85 since climatic conditions are known to impact significantly on olive production. Alternate bearing is an overall response of cropping due to yearly overlapping between two biennial cycles.86 This irregularity can actually cause fruit produced in the high yielding year to be coarse or even valueless.87 The involvement of intermediates of the cinnamic acid–lignin pathway on flower bud differentiation, rooting and callus development has been studied.85,88 The endogenous level of chlorogenic acid in olive leaves of fruit bearing trees was 3–4 times higher than the non-bearing ones.Application of chlorogenic acid decreased the amount of differentiating buds when injected prior to flower bud induction but had no effect when applied thereafter.Processing and storage The major uses of olives, namely olive oil and table olives involve extraction and/or chemical treatment of the fruit which impacts on the phenolic content of the resulting product and hence product stability and quality.89 Incorrect storage can also result in a reduction in ‘total phenols’ and other quality parameters as shown by the data of Table 5.The composition of olive fruit and olive oil exhibits some notable differences which are attributed to a series of chemical and enzymatic alterations of some substances during oil extraction. These modifications include hydrolysis of glycerides by lipases, with the formation of free fatty acids, hydrolysis of glycosides and oligosaccharides by glucosidases, oxidation of phenolic compounds by phenoloxidases and polymerisation of free phenols.90 The major phenolic compounds identified in Table 5 Variation in the quality of oils obtained from two varieties of olives stored in jute sacks (data from ref. 66, p. 19). Olive varieties Oil extraction date Organoleptic evaluation Free acidity (%) (as oleic acid) Peroxide value/mequiv. O2 kg21 trans-Hex-2-enal (ppm) Isoamyl alcohol (ppm) Total alcohols (ppm) Total phenols (ppm) DRITTA Nov. 16 7.1 0.45 7.0 279.5 4.5 45.8 578.6 Nov. 20 6.2 0.73 12.0 175.3 33.1 57.2 172.1 Nov. 23 5.4 1.21 10.5 62.4 56.0 115.2 130.8 Nov. 28 3.8 3.25 10.5 3.3 85.6 187.6 32.2 Dec. 07 3.6 7.27 17.1 1.0 96.8 256.0 — LECCINO Nov. 17 7.0 0.33 4.2 924.2 1.7 10.8 703.7 Nov. 20 6.5 0.36 11.1 450.9 7.4 24.9 484.5 Nov. 23 6.0 0.36 11.2 345.3 9.1 22.9 142.5 Nov. 28 5.1 1.24 19.1 11.8 45.0 68.5 137.9 Dec. 07 4.5 4.79 15.0 8.4 66.7 73.6 — 36R Analyst, May 1998, Vol. 123both olive fruit and virgin olive oil (see Table 2) include tyrosol, hydroxytyrosol, caffeic, p-coumaric and vanillic acids,18,91 whilst the glycosides oleuropein and verbascoside18,92 in conjunction with ligstroside, demethyloleuropein9 and the flavonoids luteolin-7-glycoside and rutin66 have been isolated from olive pulp.Ferulic, homovanillic, p-hydroxybenzoic, protocatechuic and syringic acids have also been isolated from virgin olive oil.91 Duran90 has extended the number of known phenolic compounds in virgin olive oil and has characterised some of them according to their specified role in the oil. The apparent reduction in glycosidic and flavonoid compounds in olive oil compared to olive pulp may be attributed to glycosidic modification or degradation as a result of oil extraction, which may arise due to the addition of water to the olive paste. The relative contribution of partition phenomena to the reduction has not been examined.Nevertheless, it is likely to contribute significantly, particularly in the case of the more hydrophilic phenolics. Oil production Oil production commences with the grinding of the fruit to form olive paste, which is then used for oil extraction by, for example, centrifugation, pressure or percolation. The method of oil extraction has a significant effect89 on the content of both total phenols and 1,2-diphenols.The various extraction systems differ in two important aspects, namely, the physical forces used to recover the oil, and the amount of water added to the olive paste during extraction. Oil extraction is more effective with olives of a lower water content93 Furthermore, phenolic compounds, which are critical to the organoleptic quality of olive oil, are water soluble, and so addition of water to the olive paste effectively reduces the phenolic content and quality of the oil produced.This conclusion is supported by the findings of Di Giovacchino94 who investigated the effect of the three different extraction systems on olive oil quality.Table 6 shows selected results from this investigation which found that the total phenol and o-diphenol content of oils obtained by pressing and percolation were significantly greater than that of the centrifugally extracted oils.Nevertheless, the organoleptic rating of oils obtained by the three processes was the same and hence the system of choice to ensure the highest quality oil remains a controversial issue. This can largely be attributed94 to the natural variability in the chemical composition of olive fruits. Table olives The focus in the production of table olives is the reduction of the characteristic bitterness of olive fruits.This is achieved by lye treatment, which hydrolyses the phenolic glycoside, oleuropein35, the main contributor to fruit bitterness. Brenes- Balbuena et al.18 demonstrated that the concentration of tyrosol, p-coumaric acid and vanillic acid in an experimental set of olives remained constant throughout processing whereas the concentration of caffeic acid and hydroxytyrosol declined markedly after lye treatment.This behaviour was therefore attributed to differences in chemical structure and the fact that caffeic acid and hydroxytyrosol possess an o-diphenol group. Analysis The structural diversity of the phenolics and its effect on physicochemical behaviour, such as solubility and analyte recovery, presents a challenging analytical problem. Moreover, a number of phenolic compounds are easily hydrolysed and all are relatively easily oxidised which further complicates sample handling.30,41 Sample preparation Sample preparation encompasses a series of steps ranging from exhaustive solvent extraction, filtration and concentration procedures to simple liquid–liquid extraction. Isolation of phenolic compounds from the sample matrix is a necessary prerequisite to any comprehensive analysis scheme, but it is a difficult task because the olives constitute a ‘natural’ matrix, and hence extreme care must be taken to ensure correct extraction, devoid of chemical modification, which will invariably result in artefacts.15 The precise procedure will depend on the nature of the sample (olives fruit or leaves, oil) and the desired class of phenols to be extracted.15 This accounts for the different techniques employed by Amiot et al.9 and Vlahov44 who examined total phenols and flavonoids, respectively. Similarly, Montedoro and co-workers30,37,41 have concentrated only on the simple and hydrolysable phenolic compounds present in virgin olive oil and hence have adopted an appropriate extraction procedure.Extraction of phenolics from olive oils is generally achieved by dissolution of the oil in hexane, followed by liquid–liquid extraction using various mixtures of water and alcohol in order to isolate the desired analytes from unsaturated, interferring species.32,40,95 Of the solvents examined, a methanol–water (80 + 20 v/v) mixture provided the highest recoveries of phenolics41 measured as Folin-Ciocalteu total phenols.The addition of specific lipid solvents (hexane, light petroleum, chloroform) to the oils did not enhance the phenolic concentration of the extracts. Hexane provided best selectivity in clean-up of the methanolic extract prior to HPLC.30,37,41 More recently, the versatility of solid phase extraction (SPE) has been exploited96 for the recovery of phenolics from olives. Suitable sorbents are alkylsilicas (C8 or C18)36,38 and anion exchangers.97 The oil sample was typically applied36,98 to a preconditioned Sep Pak C18 cartridge, which was then washed with a hexane–ethoxyethane mixture to remove the non-polar fraction. Phenols were then eluted with methanol, filtered, evaporated to dryness and reconstituted in water for analysis by reversed-phase chromatography (RPC). Consistent recoveries over 95% were achieved from spiked samples in contrast to the variable results with solvent extraction.The polar fraction of virgin olive oil obtained by extraction with aqueous methanol45 was fractionated into two parts (A and B) by SPE.Analysis of the two fractions showed that part A (eluted from Sep Pak C18 with methanol–water, 20 + 80) contained only simple phenols and phenolic acids whereas part B (eluted with mixtures of methanol–chloroform) had a complex nature. The two parts tested for their antioxidant Table 6 Selected quality characteristics of olive oils obtained by pressing, percolation and centrifugation94 Mini- Maxi- Determination System Average* mum mum Total phenols (gallic acid, mg l21) Pressing 158 a 111 197 Percolation 157 a 103 185 Centrifugation 121 b 87 158 o-Diphenols (caffeic acid, mg l21) Pressing 100 a 66 154 Percolation 99 a 62 149 Centrifugation 61 b 32 92 Organoleptic rating Pressing 6.9 a 6.2 7.4 Percolation 7.0 a 6.7 7.4 Centrifugation 7.0 a 6.7 7.2 * Different letters indicate significant differences at P < 0.05. Analyst, May 1998, Vol. 123 37Ractivity showed relatively high protection factors in safflower oil although part B was found to contribute more than part A to the stability of the oil. This agrees with the findings of Montedoro et al.30,41 The antioxidant activity of both fractions was related to their content of total phenols and o-diphenols although very little is known about the composition and nature of the Part B fraction. A tentative structure was assigned to one of the components in Part B using electron impact ionisation mass spectrometry EPI-MS.This paper demonstrates the potential of SPE in such studies. The versatility of SPE for preconcentration in on-line methods, coupled for example to an HPLC has not been exploited. Isolation of phenolic compounds from olive fruit is more exacting than that from olive oil. This can be attributed to the greater homogeneity and reduced enzyme content of the oil compared with the fruit. Hence, extraction of phenolics from olive fruit requires more sample handling, such as filtration to remove solid components, which therefore increases the chance of modification of the phenolics and the relative degree of error in the particular analysis.This, however, is unavoidable when dealing with natural samples. The extraction method employed by Amiot et al.9 has been used with minor modifications in several investigations.18,73 Amiot et al.9,43 concentrated their efforts on the profiling of phenolic compounds as a function of physiological development.Sample preparation entailed freeze drying and powdering the olives with the aid of liquid nitrogen. The powder was extracted twice with 80% ethanol in the presence of metabisulfite (2%) and concentrated under vacuum. Four successive light petroleum extractions of the ethanolic extract were then performed to ensure lipid and pigment removal, followed by three successive ethyl acetate washes in the presence of ammonium sulfite (20%), metaphosphoric acids (2%) and methanol (20%).The final extracts were then evaporated, and the residue dissolved in methanol for subsequent HPLC analysis. As an alternative to liquid nitrogen, the ethanol extraction step has been performed35 at 230 °C or following freeze drying.25 Nevertheless, the inclusion of an aqueous alcohol extraction and liquid–liquid fractionation typically involving ethyl acetate was universally adopted. Vlahov44 adopted a simpler approach for flavonoid analysis in which olive pulp was subjected to three successive methanol– water (80 + 20 v/v) extractions.The combined extracts were evaporated to dryness, reconstituted in glacial acetic acid–water (5 + 95 v/v) followed by centrifugation, filtration and finally HPLC analysis. The paper by V`azquez Roncero et al.99 although somewhat dated is notable for the extensive nature of the work. In this study,99 the main phenolic compounds in olive pulp were identified after extraction using acetone and methanol, and fractionation with bidimensional paper chromatography (PC).Characterisation was achieved by spectrophotometric and chromatographic methods, primarily thin-layer chromatography (TLC), using an extensive range of solvents and colorimetric reagents to distinguish between different classes of phenolics. Commonly used methods of that era involving precipitation of phenolic compounds with lead acetate, along with the use of sodium hydrogencarbonate, sodium carbonate and sodium hydroxide were avoided99 due to presumed phenolic modification and the occurrence of artefacts.Extraction procedures have generally not been subjected to rigorous quality checks and warrant closer examination to eliminate the possibility of qualitative and quantitative changes induced by the recovery procedure. Quantification Traditional methods for the determination of the phenolic component relied on measurement of total phenols or, in some instances, 1,2-diphenols89 because of their association with browning reactions.The usual approach100,101 is slow and tedious typically requiring 1 h per analysis. It involves a liquid– liquid extraction of the analytes from either the olives or, more usually, the oil into an aqueous alcohol mixture intended to isolate them from unsaturated interferents. The extraction time is an important consideration as longer times increase the possibility of oxidation of phenolic compounds unless reducing agents are added to the solvent system.An aliquot of the aqueous phase is mixed with one of a number of reagents of varying selectivity. Folin-Ciocalteu reagent is the classic reagent recommended for total phenols. An aliquot of the aqueous extract of the oil is reacted with Folin-Ciocalteau reagent in sodium carbonate solution and the blue colour formed after 15–60 min is measured100 at 725 nm. Results are expressed in terms of molar equivalents of a commonly occurring phenolic, for example, gallic acid.Folin-Ciocalteu reagent is widely used but is not specific and detects all phenolic groups in the sample extract including those found in the extractable proteins. A further disadvantage is the interference of reducing substances, such as ascorbic acid. The concentration of 1,2-diphenols, on the other hand, is determined101,102 with molybdate by measurement at 350 nm. The problem of lengthy analysis times has been overcome by the application of flow injection procedures,95 which facilitate rapid analysis with high sample throughput.Similarly, Wang et al.103 have developed a rapid procedure for determining total phenols in olive oil based on an organic-phase enzyme electrode. The method uses continuous liquid–liquid extraction to obviate the need for sample extraction and to facilitate high speed flow injection determinations of phenols in olive oil. The procedure developed by Ca�nizares et al.104 for the determination of phenols in oil used on-line coupling of a liquid–liquid extraction flow reversal system to a spectrophotometric flowthrough sensor.There are several difficulties associated with direct spectrophotometric measurement whether in flow injection or batch mode. The diversity of phenolic compounds means that selection of a reagent and/or absorbing wavelength will be a compromise although this is less of a problem where a single class of phenolics predominates.All phenols absorb radiation in the ultraviolet (Table 7) and this provides the basis for an alternative measurement of total phenols. The limited use of direct spectrophotometric measurements whether in the ultraviolet or visible region can be attributed in part to the lack of specificity of such methods. In general, they lead to an overestimation of ‘phenolic’ content. Specificity can be enhanced in direct spectrophotometric methods by derivative spectrometry.For instance, measurement based on the second derivative of the absorbance at 278 nm provided105 a rapid, direct method for determination of total Table 7 Spectral properties of various phenolic compounds (lmax) in methanol, except for anthocyanic pigments where the solvent was methanolic HCl 0.01% (data from ref. 15, p. 14). Class of compounds UV band B UV band A Visible Benzoic acids 270–280 Hydroxycinnamic acids (290–300)* 305–330 Anthocyanic pigments 270–280 (315–325)† 500–550 Flavonols 250–270 (300)* 350–380 Flavan-3-ols 270–280 Coumarins 220–230 310–350 Flavones 250–270 330–350 Flavanones, flavanonols 270–295 (300–330)* Chalcones 220–270 (300–320)* 340–390 Aurones 240–270 370–340 Isoflavones 245–270 300–340 * Shoulder. † In the case of acylation by hydroxycinnamic acids. 38R Analyst, May 1998, Vol. 123phenols. Catechol was the most appropriate reference standard. Chromatographic methods The need for profiling and identifying individual phenolic compounds has seen traditional methods based on colorimetry replaced by chromatographic analyses.Bate-Smith106 pioneered the identification of plant phenolics through the application of PC,107 and in the 1950s and 1960s, many paper chromatographic methods were developed for such purposes. PC was subsequently superseded by TLC. The usual advantages of TLC, namely speed and an open-bed technique are realised in phenolic analyses. Similarly, the versatility of the technique is evidenced in the extensive array of solvent systems available which can be exploited for specific class analyses.Detection of phenolics may be achieved by viewing the chromatogram under UV light both before and after exposure to ammonia fumes, which often changes the colour107 of their fluorescence. Many phenolic compounds also give characteristic colours when treated with diazotised p-nitroaniline, sulfanilic acid, p-toluenesulfonic acid plus vanillin and heating30 or iron(iii) chloride. 108 Contrary to the relative ease and adaptability of TLC, the use of this technique for phenolic characterisation in olive matrices is limited. Ragazzi and Veronese109 have developed a method for phenolic quantification using UV spectrophotometry after analyte separation using either silica gel or cellulose thin-layers. Quantification was achieved using Folin-Ciocalteau reagent. The main phenols with the exception of oleuropein occurring in olive vegetation water, namely catechol, 4-methylcatechol, tyrosol and hydroxytyrosol have been detected108 by reversedphase TLC (RP-TLC) whereas high-performance TLon silica layers (Si-HPTLC) yielded the detection of only tyrosol and hydroxytyrosol.Confirmation of the identities of the phenolic compounds was obtained by RP-TLC and Si-HPTLC analysis of the acetylated organic extracts of the water using the more stable acetyl derivatives of the phenols as standards. The flavonoid content of olive leaves was studied21 by TLC on silica gel and SIL layers and recording UV spectra of the isolated spots directly on the layer.SIL is a hydrophobic layer and a reversed-phase mechanism operates while silica gel gives normal phase behaviour. From comparison of the chromatographic behaviour and UV spectra three flavonoid glycosides (quercitrin, rutin and luteolin-7-glycoside), one flavonoid (aglycone), luteolin, and chlorogenic acid were identified in the leaves. Flavonoid glycosides were distinguished from the flavonoid aglycone by two-dimensional TLC.Montedoro et al.30,41 have used silica gel TLC for the preliminary isolation of phenolic compounds from virgin olive oil. Four different mobile phases were used covering a range of polarities and selectivities. After separation the phenolics were extracted from the silica layers for analysis by HPLC and UV spectroscopy. The limited application of gas chromatography to the separation of olive phenolics (see Table 8) can be attributed114 to their polarity and limited volatility. Hence, a derivatisation step is usually mandatory, thermal decomposition may occur during their elution and the higher molecular mass phenolics cannot be chromatographed.Nevertheless, the excellent resolving power and detection capabilities of GC warrant consideration. Indeed, many useful separations have been achieved and this has been facilitated by the availability of inert open tubular columns.This is seen in the elution of phenolic acids110 from olive leaves and roots using a 30 m SPB-1 column programmed from 138 to 150 °C. Acid hydrolysis of the plant tissues was employed to obtain free phenolic acids from conjugated forms such as the O-glycosides. The free phenolic acids were recovered from the leaves and roots by extraction with ethyl acetate and converted to the corresponding trimethylsilyl derivatives prior to GC. Experimental conditions were chosen to eliminate interference by sugars and flavonoids.The most abundant phenolic acids were salicylic, cinnamic, o-coumaric and ferulic acids. Qualitative differences in the distribution of the phenolic acids were demonstrated between cultivars and within different tissues of the same cultivar notably for shikimic and syringic acids. GC–MS is now well established as a routine technique carried out with either electron impact ionisation (EI) or chemical ionisation (CI) sources, since these are appropriate for the introduction of volatile compounds.However, because of limited volatility, analysis of phenolic compounds and their glycosides, in particular, by GC and thus GC-MS has not generally found favour. Nevertheless, Angerosa et al.92 have shown GC–MS to be an effective tool for phenol identification after extraction from olive oil with methanol and derivatisation with bis(trimethylsilyl)trifluoracetamide. Peaks in the mass spectra at m/z 192 or at m/z 280, related only to tyrosol and hydroxytyrosol, were attributed to a McLafferty rearrangement of linked phenols, and were useful for assigning the phenolic nature to minor components. Proposed structures for these linked phenolic compounds and their hypothesised interconversion are shown in Fig. 3. In a later paper, the advantages of chemical ionisation with ammonia for providing molecular masses of the aglycones from the glycosides, ligstroside, decarbomethoxyoleuropein and oleuropein were demonstrated. 111 The phenolic components of wine have been extracted and separated115 as trimethylsilyl derivatives on a DB- 5HT capillary column using MS detection with one target and two qualifying ions for each compound in a total run time of 26 min. Resolution of all 15 phenolic compounds was excellent and the method should be appropriate for phenolics in olives following suitable extraction. Liquid chromatography. Phenolic extracts from olive oil have been fractionated by classical low-pressure column chromatography on, for example, Sephadex LH20.30 Alternatively, preliminary fractionation can be achieved on ion exchange columns as demonstrated116 for carboxylic acid phenolics and non-carboxylic acid phenolics of maize.The poor efficiency of such separations has favoured development of RPC, which currently represents the most popular and reliable technique for phenolic analysis. Compound elution is typical of RPC, that is, polar compounds (e.g., phenolic acids) elute first, followed by those of decreasing polarity.Thus, the typical elution pattern19,35,113,117 is hydroxytyrosol < tyrosol < vanillic acid, caffeic acid < p-coumaric acid < elenolic acid < verbascoside < rutin < luteolin-7-glucoside < oleuropein < ligstroside. In one of the early reports117 on the RPC of phenolic compounds, different mixtures of acetic acid, water and methanol were used to separate members of several classes of phenols, and the effects of organic modifier on selectivity were deduced.Since then, numerous mobile phases have been employed (Table 8) with different modifiers (usually methanol, acetonitrile or tetrahydrofuran), acids (acetic or formic acid) and/or salts (ammonium phosphate). Gradient elution has usually been mandatory in recognition of the complexity of the phenolic profile although isocratic elution has been successful for particular applications.38 In some instances, the success of isocratic elution can be attributed to selectivity effects of one or more components (e.g., acetonitrile) of the mobile phase.38 The most popular stationary phases have involved C1832,40 chemistry.In a typical application, phenolics were recovered from olive fruit35 by extraction in the presence of metabisulfite. After suitable clean-up, the extract was chromatographed on a Spherisorb ODS-2 column using gradient elution with acetonitrile–water (containing phosphoric acid).Eluted species were identified from their retention times and absorption spectra in the 280–380 nm range. Analyst, May 1998, Vol. 123 39RTable 8 Conditions used for the analysis of phenolic compounds in olives Sample Method Column Mobile phase Detection Comment Ref. Fruit Counter-current chromatography 300 columns; 400 3 2 mm Chloroform, methanol, water Fraction collection Isolation of oleuropein and ligstroside 34 Fruit RPC 300 3 4 mm Micropak MCH-5 Gradient; acetonitrile, water, phosphoric acid 280 nm, 340 nm Oleuropein; verbascoside, rutin, luteolin-7-glucoside 9 Fruit RPC 300 3 4 mm Micropak MCH-5 Gradient; acetonitrile, water, phosphoric acid 280 nm, 340 nm Effect of black maturation on oleuropein, demethyloleuropein and elenolic acid glucoside 43 Oil RPC Reverse phase (no further details provided) Ternary gradient 280 nm Effect of ripening on formation of tyrosol and hydroxytyrosol by hydrolysis of higher molecular mass phenols 42 Leaves TLC Silica gel; SIL C18-50 Various Densitometer Determination of flavonoids and flavonoid glycosides 21 Oil RPC 150 3 4.6 mm C18 Gradient; 2% acetic acid in water, methanol 239 nm, 278 nm Elenolic acid; phenolic acids 41 Oil TLC RPC Silica gel 150 3 4.6 mm Erbasil C18 Various Gradient; 2% acetic acid in water, methanol Various spray reagents 239 nm, 278 nm Characterisation of hydrolysable phenolic fraction of oil (oleuropein aglycone, elenolic acid) 30 Oil RPC 250 3 4.6 mm Spherisorb ODS2 Gradient: acetic acid, methanol, water 280 nm Effect of hydroxytyrosol and tyrosol on stability of oil 100 Oil RPC 250 3 4.6 mm Spherisorb ODS2 Gradient: methanol, water, acetic acid 280 nm Use of UV detection 40 Fruit RPC 250 3 4.0 mm Spherisorb ODS2 Gradient: water, acetonitrile, phosphoric acid DAD Effect of lye treatment on hydroxytyrosol, verbascoside, tyrosol, vanillic acid, p-coumaric acid, oleuropein 18 Leaves RPC 250 3 4.6 mm Ultrasphere ODS Gradient; acetonitrile, water, tetrahydrofuran, phosphoric acid 280 nm, 340 nm Antioxidant activities of oleuropein and flavonoids 33 Fruit RPC 100 3 4.6 mm Microspher C18 Gradient; methanol, water, acetic acid 520 nm Effect of maturation on flavonoid content 44 Oil RPC 250 3 4.6 mm m Bondapak C18 Isocratic; acetonitrile, water, acetic acid Amperometric Use of electrochemical detection 38 Oil RPC 500 3 9.4 mm Partisil 10 ODS2 Gradient; water, methanol, acetic acid 278 nm Isolation of four new phenolic compounds 37 Fruit RPC 250 3 4 mm Spherisorb ODS2 Gradient; water, acetonitrile, phosphoric acid 280 nm Effect of cultivar and processing on levels of oleuropein, verbascoside, and luteolin-7-glucoside; hydroxytyrosol increased due to hydrolysis of major phenolic compounds in brines 19 Leaves, roots GC 30 m 3 0.32 mm SPB-1 Helium FID Phenolic acid composition 110 Oil GC–MS 25 m 3 0.32 mm SE-54 Helium MS Phenol identification 92 Oil RPC 250 3 4.6 mm Spherical Resolve C18 Gradient; water, methanol, acetic acid Amperometric Use of amperometric detection 36 Oil RPC 250 3 4.0 mm Lichrosorb RP18 Gradient; water, acetic acid, methanol, acetonitrile 280 nm Effect of anti-Dacus treatment on hydroxytyrosol, tyrosol, vanillic acid and p-coumaric acid 32 Fruit RPC 250 3 4 mm Spherisorb ODS2 Gradient; water, acetonitrile, phosphoric acid 280 nm Changes in phenolic compounds during olive processing 35 Fruit RPC 250 3 4 mm Lichrospher RP18 Gradient; acetic acid, methanol DAD Effect of altitude on phenolic acid content of olives 17 Oil GC–MS 30 m 3 0.25 mm DB5 Helium CI–MS Characterisation of phenolic compounds 111 Oil RPC 250 3 4.6 mm Spherisorb ODS Gradient; methanol, water 280 nm Antioxidant activity of various fractions; some mass spectral data 45 continued next page 40R Analyst, May 1998, Vol. 123O COOCH3 O HO HO O O O O HO HO OH O O O HO HO O O The limited availability of suitable reference standards is a problem which has been overcome, in part, by synthesis92,111,113 of the relevant compounds.Vlahov44 has reported the only detailed examination of the change in flavonoids during olive fruit maturation. Based on RPC with detection at 520 nm (anthocyanins) or 350 nm (flavones and flavonols), different olive varieties were characterised by their flavonoid profiles. Numerous papers demonstrate the power of RPC for analysis of the phenolic fraction of olives. For example, with RPC and NMR Montedoro et al.30,37,41 separated and identified some aglycone derivatives present as the dialdehydic forms of elenolic acid linked to both hydroxytyrosol and tyrosol in the olive oil.In an atypical study113 involving low wavelength detection at 225 nm, phenolics were separated by RPC using a stepwise gradient of sulfuric acid and acetonitrile. The response of a range of simple and complex phenols was 3–14 fold higher at 225 nm than at the more usual detection wavelength of 280 nm. Apart from tyrosol, hydroxytyrosol, oleuropein aglycone and elenolic acid, dialdehydic derivatives of oleuropein and ligstroside were identified in the chromatograms. Further signals of unknown, but possibly phenolic substances, were also detected at the lower wavelength.Data for one of these peaks were consistent with the elution of an oleuropein derivative previously assigned118 as deacetoxyoleuropein aglycone. Nevertheless, problems associated with high background absorption of typical mobile phases in RPC have limited the use of low wavelength detection.This study is also interesting for its use of SPE on C8 cartridges for recovery of phenolics and for the systematic investigation of stationary phases for the analytical separation. Phases examined for this purpose were ODS2, ODS1, C8 and phenyl phases and whilst all showed similar retention behaviour the best separation was achieved on ODS2 columns. Phenolics were extracted31 from oil samples obtained from olives that had reached different degrees of ripeness and that had been affected by Dacus oleae infestation differently. RPC of the extracts using a quaternary mobile phase showed 23 significant peaks in the chromatograms.Such data are ideally suited to chemometric analysis and partial least squares regression produced models that showed a significant correlation between the phenolic composition of the oil and conditions of the olives sampled. In particular, the first principal component reflected the o-diphenol content of the oil and was directly linked with the state of health of the olives.Moreover, prediction of the shelf life of the oil was possible. Detection. Detection in RPC is typically based on measurement of UV absorption. No single wavelength is ideal for all classes of phenolics since they display absorbance maxima at distinctly different wavelengths (Table 7). Indeed, there are significant differences in absorption maxima and molar absorptivities40 of even the major phenolics identified in olives.This creates problems in quantification as discussed by Tsimidou et al.40 who classified the various phenolics into four groups and used a single calibration standard for the members of each group. The results suggest that in cases of unidentified phenols, it is preferable to report data as peak areas rather than to assign concentrations42 using an arbitrary reference. On the other hand, these different spectral characteristics can be exploited (Fig. 4) to provide useful qualitative information about an eluted species and in the ideal case enable selective detection. The most commonly used wavelength has been 280 nm which represents a suitable compromise,32,40,116 although detection at other wavelengths including 340 nm9 has been applied. The choice of detection wavelength will invariably depend on the desired class of phenolics to be investigated. Hence, the absorption maximum around 340–350 nm has been used,33,44 for example, for flavonoid analysis whereas elenolic acid glucoside from the hydrolysis of oleuropein was detected35 at 240 nm.Dual wavelength measurement at 278 and 239 nm provided some interesting differences in the resulting chromatograms as might be anticipated.41 Elenolic acid was identified in the samples from additional information in the chromatogram at 239 nm. Such measurements are conveniently performed with a photodiode array detector.The extensive use of photodiode array detection can be attributed to the ability to collect on-line spectra without using stop–flow techniques. The UV spectra of phenolic compounds are particularly informative (Table 7) providing considerable structural information. Furthermore, spectra of eluting peaks obtained at, for example, the apex and both inflexion points of the peak can be compared and used as an indicator of purity. Nevertheless, in most instances, diode array detection (DAD) has been employed for fixed wavelength or, at most, dual wavelength detection.Hence, the outstanding capabilities of this mode of detection have not been realised. RPC using amperometric detection has also been used successfully36,98 for the quantitative determination of phenolic Table 8 Continued— Leaves LC–API–MS 300 3 4.6 mm C18 Gradient; acetonitrile, water, formic acid API-MS Characterisation of phenolic glucosides 46 Oil RPC 250 3 4.0 mm Lichrosorb C18 Gradient; water, acetic acid, methanol, acetonitrile DAD Chemometric analysis 31 Water LC–APCI–MS 150 3 2 mm Supelcosil LC-18 Gradient; methanol, water, formic acid 280 nm; APCI–MS APCI mass spectra 112 Oil RPC Various including ODS-1, ODS-2, phenyl and C-8 Stepwise gradient; sulfuric acid, acetonitrile 225 nm Low wavelength detection 113 Fig. 3 Chemical structure of linked phenolic compounds and their hypothesised interconversion.92 Analyst, May 1998, Vol. 123 41Rcompounds in virgin olive oils.The detector system employed a dual electrode detector in the parallel configuration, operating at +0.5 and 1.0 V vs. Ag/AgCl. The voltammetric behaviour was useful in assigning peak identity. For example, o- and pdiphenols were easily recognised by their facile oxidation to oand p-quinones, respectively, in the range from +0.5 to +0.6 V. On the reverse scan, a cathodic wave due to the reduction of the quinone species formed was also observed for compounds having the catechol moiety.The authors assigned an antioxidant activity to the phenolic species in olive oil based on their measured oxidation potential. Other investigators have conducted studies using cyclic voltammetry to optimise amperometric detection conditions38 and to provide selective as well as sensitive detection of the major phenolic compounds in olive oil. A number of techniques which have the potential to improve sensitivity and/or selectivity of phenolic analyses appear to have been ignored.Thus, post-column derivatisation offers a number of advantages including enhanced selectivity. Fluorescence detection is an obvious means of improving both sensitivity and selectivity. It is interesting that one of the earliest papers on HPLC of phenolic compounds119 employed this means of detection as an adjunct to conventional UV detection. Stop– flow scans were used to obtain excitation and emission spectra of the eluted species. This early paper demonstrated the complementary nature of the two methods of detection.The limited stability and light sensitivity of several phenolics were noted and should serve as a warning. The on-line coupling of RPC and MS is of enormous potential because the selectivity can then be tuned in an optimal way. Classical mass spectrometric gas phase ionisation techniques, such as EI and CI, are generally less suitable for polar, non-volatile compounds such as the phenolics. The power of atmospheric pressure ionisation (API) methods, such as electrospray (ES), as alternative, highly sensitive soft ionisation techniques for investigation of polar, non-volatile and thermolabile molecules has been demonstrated (Fig. 5). API procedures overcome the lack of analyte volatility by direct formation or emission of ions from the surface of a condensed phase. Hence, they eliminate the need for neutral molecule volatilisation prior to ionisation and generally minimise thermal degradation of the molecular species.Atmospheric pressure chemical ionisation (APCI) is a development of ES in which a combination of a heated capillary and a corona discharge is used to promote the formation of ions from the nebulised sample. As the name implies, APCI involves gas phase ion–molecule reactions which cause the chemical ionisation of analyte molecules under atmospheric pressure conditions. Aramend�ýa et al.112 reported the LC–APCI-MS of phenolics in olive mill wastewater. Analytes were separated on a C18 phase by gradient elution with methanol–water containing formic acid.Mass spectral conditions were optimised by direct infusion of standards in flow injection mode into the APCI source. The study was restricted to negative-ion mode with detection limits in total ion current mode ranging from 0.5 to 500 ng. These detection limits were about 20 times better when working in selected ion monitoring mode and monitoring the [M–H]2 ion.Mass spectra were recorded with soft (215 V) and strong (250 V) voltages applied at the ion source of the mass spectrometer. With the smaller voltages, deprotonated molecular species [M–H]2 were the major ions observed in the mass spectra with the appearance of very few fragment ions which Fig. 4 Chromatograms showing the effect of detection wavelength in RPC. Phenolics were extracted from freeze-dried green olives by SPE and separated on a Varian C18 column (150 mm) using gradient elution.Detection wavelength (a) 280 nm or (b) 340 nm. Fig. 5 Total ion chromatograms in (a) positive- and (b) negative-ion mode obtained by ESI-LC–MS using gradient elution and RPC. The peak eluting at 16.5 min is confirmed as oleuropein. Samples were obtained as for Fig. 4. 42R Analyst, May 1998, Vol. 123were all of low intensity. The presence of substantial fragmentation from collisionally induced dissociation processes, which became evident on increasing the voltage applied at the source (extraction and cone) voltages, gave structural information about the molecules.Structures were assigned to major eluent cluster ions from methanol–water–formic acid mixtures occurring at m/z 91, 113, 137, 159, 181 and 183. Ionspray (or pneumatically assisted electrospray) has been applied46 to the identification of the phenolic glucoside content of olive leaves following extraction with methanol and partitioning in acetonitrile–hexane. Structural data on oleuropein and ligstroside were obtained from the positive-ion spectra.Moreover, the presence of a disaccharide containing the hydroxytyrosol moiety was confirmed. The bioactivity of the phenolics (P) is exerted by supramolecular formation, for example, between the phenol and sensorial receptors (SR) on the tongue or globular and prolinerich mucoproteins (MP) or other food components (FC). The specific interaction among P, SR, MP and FC may involve absorption and desorption equilibria with formation of charge transfer host–guest aggregates.FAB-MS has been used120 to study such supramolecular formations between hydroxytyrosol and caffeine or the dipeptide, Asp–Phe, as protein models. The data demonstrated a preferential molecular recognition site provided by caffeine, the biomimetic model of proline-rich mucoproteins. The complexity of the biochemical processes controlling phenolic metabolism have been shown39 in a 1H NMR study of phenolics in three cultivars and their changes with fruit development.The major phenolic compounds, common to all three cultivars, were identified from chemical shifts, peak multiplicities and scalar correlations in two-dimensional and selective excitation experiments on aqueous extracts (olive vegetation water) as tyrosol, 4-hydroxyphenylethanol glucoside, and oleuropein. Considerable differences in the content of these compounds occurred in the fruits during growth and maturation of the drupe.In contrast, the glucoside, cornoside was detected in only two of the cultivars. Possible metabolic pathways leading to cornoside and halleridone (not a phenol) were discussed. Of most interest, is the discrepancy noted by Limiroli et al.39 between the limited number of substances found in the olive vegetation waters and the large number of phenolic compounds cited in the literature as occurring in olives. References 1 Kiritsakis, A., and Markakis, P., Adv.Food Res., 1987, 31, 451. 2 Bonazzi, M., Olivae, 1997, 65, 16. 3 Parras Rosa, M., Olivae, 1996, 63, 24. 4 Navarro, M. D., Periago, J. L., Pita, M. L., and Hortelano, P., Lipids, 1994, 29, 845. 5 Charbonnier, A. (1982). Main conclusions drawn from the international symposium on the recent medical research on the value of olive oil to health. In international symposium on the recent medical research on the value of olive oil to health, Paris. November 17, pp. 1–4. 6 Kuwajima, H., Uemura, T., Takaishi, K., Inoue, K., and Inouye, H., Phytochemistry, 1988, 27, 1757. 7 Gariboldi, P., Jommi, G., and Verotta, L., Phytochemistry, 1986, 25, 865. 8 Teissedre, P. L., Waterhouse, A. L., Walzem, R. L., German, J. B., Frankel, E. N., Ebeler, S. E., and Clifford, A. J., Bull. O.I.V., 1996, 69(781–782), 251. 9 Amiot, M.-J., Fleuriet, A., and Macheix, J.-J., J Agric. Food Chem., 1986, 34, 823. 10 Perrin, J. L., Rev. Fr. Corps Gras, 1992, 39, 25. 11 Lattanzio, V., Cardinali, A., and Palmieri, S., Ital. J.Food Sci., 1994, 6, 3. 12 Robards, K. and Antolovich, M., Analyst, 1997, 122, 11R, and references cited therein. 13 Leibovitz, B. E., and Mueller, J. A., J. Optimal Nutrition, 1993, 2, 17. 14 Huang, M. T., and Ferraro, T., ACS Symp. Ser., 1992, 507, 8. 15 Macheix, J. J., Fleuriet, A., and Billot, J., Fruit Phenolics, CRC, Boca Raton, FL, 1990. 16 Movsumov, I. S., Aliev, A. M., and Tagieva, Z. D., Farmatsiya (Moscow), 1987, 36, 32. 17 Mousa, Y.M., Gerasopoulos, D., Metzidakis, I., and Kiritsakis, A., J. Sci. Food Agric., 1996, 71, 345. 18 Brenes-Balbuena, M., Garcia, P., and Garrido, A., J. Agric. Food Chem., 1992, 40, 1192. 19 Brenes, M., Garcia, P., Duran, M. C., and Garrido, A., J. Food Sci., 1992, 58, 347. 20 Nergiz, C., and Unal, K., Food Chem., 1991, 39, 237. 21 Heimler, D., Pieroni, A., Tattini, M., and Cimato, A., Chromatographia, 1992, 33, 369. 22 Balice, V., and Cera, O., Grasas Aceites (Seville 1984, 35, 178. 23 Duran, R.M., Cabello, R. L., Gutierrez, V. R., Fiestas, P., and V`azquez Roncero, A., Grasas Aceites (Seville), 1994, 45, 332. 24 Lo Scalzo, R., and Scarpati, M. L., J. Nat. Prod., 1993, 56, 621. 25 Baldi, A., Romani, A., Mulinacci, N., Alberti, M. B., and Vincieri, F. F., Bull. Liaison-Groupe Polyphenols, 1992, 16(Pt. 2), 60. 26 Baldi, A., Romani, A., Tatti, S., Mulinacci, N., and Vincieri, F. F., Colloq.-Inst. Natl. Rech. Agron., 1995, 69, 269. 27 Vacca, V., Fenu, P., Franco, M.A., and Sferlazzo, G., Riv. Ital. Sostanze Grasse, 1993, 70, 595. 28 Poiana, M., Giuffre, A. M., Giuffre, F., Modafferi, V., Neri, A., Mincione, B., and Taccone, P. L., Riv. Ital. Sostanze Grasse, 1997, 74, 59. 29 Mincione, B., Poiana, M., Giuffre, A. M., Modafferi, V., and Giuffre, F., Riv. Ital. Sostanze Grasse, 1996, 73, 245. 30 Montedoro, G., Servili, M., Baldioli, M., and Miniati, E., J. Agric. Food Chem., 1992, 40, 1577. 31 Evangelisti, F., Zunin, P., Tiscornia, E., Petacchi, R., Drava, G., and Lanteri, S., J. Am.Oil Chem. Soc., 1997, 74, 1017. 32 Zunin, P., Evangelisti, F., Pagano, M. A., and Tiscornia, E., Riv Ital Sostanze Grasse, 1995, 72, 55. 33 Le Tutour, B., and Guedon, D., Phytochemistry, 1992, 31, 1173. 34 Kubo, I., and Matsumoto, A., J. Agric. Food Chem., 1984, 32, 687. 35 Brenes, M., Rejano, L., Garcia, P., Sanchez, A. H., and Garrido, A., J. Agric. Food Chem., 1995, 43, 2702. 36 Mannino, S., Cosio, M. S., and Bertuccioli, M., Ital.J. Food Sci., Spec Issue, 1995, 150. 37 Montedoro, G. F., Servili, M., Baldioli, M., Selvaggini, R., Miniati, E., and Macchioni, A., J. Agric. Food Chem., 1993, 41, 2228. 38 Akasbi, M., Shoeman, D. W., and Csallany, A. S., J. Am. Oil Chem. Soc., 1993, 70, 367. 39 Limiroli, R., Consonni, R., Ranalli, A., Bianchi, G., and Zetta, L., J. Agric. Food Chem., 1996, 44, 2040 and references cited therein. 40 Tsimidou, M., Papadopoulos, G., and Boskou, D., Food Chem., 1992, 44, 53. 41 Montedoro, G., Servili, M., Baldioli, M., and Miniati, E., J. Agric. Food Chem., 1992, 40, 1571. 42 Cimato, A., Mattei, A., and Osti, M., Acta Hort., 1990, 286, 453. 43 Amiot, M.-J., Fleuriet, A., and Macheix, J-J., Phytochemistry, 1989, 28, 67. 44 Vlahov, G., J. Sci. Food Agric., 1992, 58, 157. 45 Litridou, M., Linssen, J., Schols, H., Bergmans, M., Posthumus, M., Tsimidou, M., and Boskou, D., J. Sci. Food Agric., 1997, 74, 169. 46 De Nino, A., Lombardo, N., Perri, E., Procopio, A., Raffaelli, A., and Sindona, G., J.Mass Spectrom., 1997, 32, 533. 47 Tassou, C. C., and Nychas, G. J. E., J. Food Prot., 1994, 57, 120. 48 Muller, K. O., Rec. Adv. Bot., 1961, 1, 396. 49 Chowdhury, B., Bhattacharyy, D., and Mukhopadhyay, S., Biomed. Lett., 1996, 54, 45. 50 Ruiz-Barba, J. L., Garrido-Fernandez, A., and Jimenez-Diaz, R., Lett. Appl. Microbiol., 1991, 12, 65. 51 Matern, U., and Grimmig, B., Acta Hort., 1994, 381, 448. 52 Teissedre, P. L., Frankel, E.N., Waterhouse, A. L., Peleg, H., and German, J. B., J. Sci. Food Agric., 1996, 70, 55. 53 Plant Flavonoids in Biology and Medicine. II: Biochemical, Cellular and Medicinal Properties, ed. Cody, V., Middleton, E., Harborne, J. B., and Beretz, A., Prog. Clin. Biol. Res., 280, Alan R. Liss, Inc., New York, 1988. 54 Cuvelier, M-E., Berset, C., and Richard, H., J. Agric. Food Chem., 1994, 42, 665. Analyst, May 1998, Vol. 123 43R55 Dziedzic, S. Z., and Hudson, B.J. F., Food Chem., 1984, 14, 45. 56 Houlihan, C. M., Ho, C-T., and Chang, S. S., J. Am. Oil Chem. Soc., 1984, 61, 1036. 57 Onyeneho, S. N., and Hettiarachchy, N. S., J. Agric. Food Chem., 1992, 40, 1496. 58 Pearson, D., Frankel, N., Aeschbach, R., and German, J. B., J. Agric. Food Chem., 1997, 45, 578. 59 Baldioli, M., Servili, M., Perretti, G., and Montedoro, G. F., J. Am. Oil Chem. Soc., 1996, 73, 1589. 60 Papadopoulos, G., and Boskou, D., J. Am. Oil Chem. Soc., 1991, 68, 669. 61 Castera-Rossignol, A., and Bosque, F., Ol., Corps Gras, Lipides, 1994, 1, 131. 62 Marinova, E. M., and Yanishlieva, N., Wl. Food Chem., 1996, 56, 139. 63 Satue, M. T., Huang, S.-W., and Frankel, E. N., J. Am. Oil Chem. Soc., 1995, 72, 1131. 64 Madsen, H. L., Nielsen, B. R., Bertelsen, G., and Skibsted, L. H., Food Chem., 1996, 57, 331. 65 Puchades, R., Suescun, A., and Maquieira, A., J. Sci. Food Agric., 1994, 66, 473. 66 Boskou, D., Olive Oil; Chemistry and Technology. AOCS Press, Champaign, IL, 1996. 67 Sanchez Saez, J. J., Herce Garraleta, M. D., and Balea Otero, T., Anal. Chim. Acta, 1991, 247, 295. 68 Mayer, A. M., and Harel, E., Annu. Proc. Phytochem. Soc. Eur., 1981, 19 (Rec. Adv. Biochem. Fruits Veg.), pp. 161–180. 69 Brenes, M., Romero, C., Garcia, P., and Garrido, A., J. Sci. Food Agric., 1995, 67, 35. 70 Garcia, P., Romero, C., Brenes, M., and Garrido, A., J. Agric. Food Chem., 1996, 44, 2101. 71 Garcia, P., Brenes, M., Vattan, T., and Garrido, A., J.Sci. Food Agric., 1992, 60, 327. 72 Sciancalepore, V., J. Food Sci., 1985, 50, 1194. 73 Goupy, P., Fleuriet, A., Amiot, M.-J., and Macheix, J. J., J. Agric. Food Chem., 1991, 39, 92. 74 Goupy, P., and Fleuriet, A., Bull. Liaison - Groupe Polyphenols, 1986, 13, 455. 75 Raina, B. L. in Handbook of fruit science and technology; Production, composition, storage and processing, ed. Salunkhe, D. K., and Kadam, S. S., Marcel Dekker, New York, 1995. 76 Kiritsakis, A. K., Olive Oil, AOCS Press, Champaign, IL, 1990. 77 Aparicio, R., Ferrero, L., and Alonso, V., Anal. Chim. Acta, 1994, 292, 235. 78 Tous, J., and Romero, A., Acta Hort., 1994, 356, 323. 79 Pannelli, G., Servili, M., Selvaggini, R., Baldioli, M., and Montedoro, G. F., Acta Hort., 1994, 356, 239. 80 Maestro Dur�an, R., Acta Hort., 1990, 286, 441. 81 Deidda, P., Nieddu, G., Spano, D., Bandino, G., Orru, V., Solinas, M., and Serraiocco, A., Acta Hort., 1994, 356, 354. 82 Alessandri, S., Cimato, A., Mattei, A., and Modi, G., Acta Hort., 1994, 356, 233. 83 Garcia, J. M., Seller, S., and Perez-Camino, M. C., J. Agric. Food Chem., 1996, 44, 3516. 84 Shulman, Y., and Lavee, S., Plant Physiol., 1976, 57, 490. 85 Lavee, S., and Avidan, N., Acta Hort., 1994, 356, 143. 86 Rallo, L., Torreno, P., and Alvarado, J. A. V., Acta Hort., 1994, 356, 127. 87 Barone, E., Gullo, G., Zappia, R., and Inglese, P., J. Hort. Sci., 1994, 69, 67. 88 Lavee, S., Avidan, N., and Pierik, R. L. M., Acta Hort., 1994, 381. 89 Nergiz, C., and Unal, K., J. Sci. Food Agric., 1991, 56, 79. 90 Duran, R. M., Acta Hortic, 1990, 286, 441. 91 Nergiz, C., Riv. Ital. Sostanze Grasse, 1991, 68, 553. 92 Angerosa, F., d’Alessandro, N., Konstantinou, P., and Di Giacinto, L., J. Agric. Food Chem., 1995, 43, 1802. 93 Di Giovacchino, L., Solinas, M., and Miccoli, M., J. Am. Oil Chem. Soc., 1994, 71, 1189. 94 Di Giovacchino, L., Olivae, 1996, 63, 52. 95 Mesa, J. A. G., Linares, P., Leque de Castro, M. D., and Valcarcel, M., Anal. Chim. Acta, 1990, 235, 441. 96 Papadopoulos, G. K., and Tsimidou, M., Bull. Liaison - Groupe Polyphenols, 1992, 16(Pt. 2), 192. 97 Andreoni, N., and Fiorentini, R., Riv. Ital. Sostanze Grasse, 1995, 72, 163. 98 Mannino, S., Cosio, M. S., and Bertuccioli, M., Ital. J. Food Sci., 1993, 5, 363. 99 V`azquez Roncero, A., Graciani Constante, E., and Maestro Duran, R., Grasas y Aceites, 1974, 25, 269. 100 Tsimidou, M., Papadopoulos, G., and Boskou, D., Food Chem., 1992, 44, 141. 101 Gutfinger, T., J. Am. Oil Chem. Soc., 1981, 58, 966. 102 Nergiz, C., Int. J. Food Sci. Technol., 1993, 28, 461. 103 Wang, J., Reviejo, A. J., and Mannino, S., Anal. Lett., 1992, 25, 1399. 104 Ca�nizares, M. P., Tena, M. T., and Luque de Castro, M. D., Anal. Chim. Acta, 1996, 323, 55. 105 Montano Asquerino, A., Rejano Navarro, L., and Sanchez Gomez, A. H., Grasas Aceites (Seville), 1985, 36, 274. 106 Bate-Smith, E. C., Nature (London), 1969, 18, 241. 107 Walker, J. R. L., The Biology of Plant Phenolics, Edward Arnold, London, 1975. 108 Capasso, R., Evidente, A., and Scognamiglio, F., Phytochem. Anal., 1992, 3, 270. 109 Ragazzi, E., and Veronese, G., J. Chromatogr, 1973, 77, 369. 110 Heimler, D., and Pieroni, A., Chromatographia, 1994, 38, 475. 111 Angerosa, F., dAAlessandro, N., Corana, F., and Mellerio, G., J. Chromatogr., A, 1996, 736, 195. 112 Aramend�ýa, M. A., Bor�au, V., Garc

ISSN:0003-2654    

DOI:10.1039/a708920a

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Speciation of organometal and organohalogen compounds in relation to global environmental pollution† Freddy C. Adams*a, Monika Heisterkampa, Jean-Pierre Candelonea, Frank Laturnus‡a, Katya van de Veldeb and Claude F. Boutronb a Department of Chemistry, University of Antwerp (UIA), Universiteitsplein 1, B-2610 Wilrijk, Belgium b Laboratoire de Glaciologie et G�eophysique de l’Environnement du CNRS, 54 rue Moli`ere, Domaine Universitaire, 38402 Saint Martin d’H`eres, Cedex, BP 96, France Speciation of the elements emitted into the atmosphere plays an important role in their long range transport over the globe and their eventual pollution of remote environments.This paper describes recent results of our laboratory in: (1) organolead determinations in archives of snow and ice in Greenland and the Mont Blanc region in Western Europe. Speciation analysis together with the determination of total inorganic lead gives a clear indication on the extent of global pollution when compounds are transported over long distances in the atmosphere; and (2) determinations of volatile halocarbons in macroalgae whose transport in the stratosphere could interfere with the ozone destruction process.It is shown that the analytical methodology for the determination of a number of species is now well enough developed to allow ultra-sensitive and reliable measurements in samples collected in the remote environment and to derive interesting conclusions on long range transport processes. Keywords: Organolead compounds; volatile organohalogen compounds; speciation; polar snow; macroalgae; Antarctic; environmental pollution With the growing concern about pollution of the global environment, interest has been expanded from local pollution studies to studies of environmental contaminants in remote regions of our planet.In this context, long range transport of air pollutants is of special concern. Concentrations of trace elements and organic compounds have been measured in many studies in air samples collected at remote stations.1,2 Results showed that the background composition had a large spatial variation in elemental composition and during cases of long range transported air pollution (episodes), concentrations of air pollutants as heavy metals, sulfur compounds and organic compounds were shown to be several times higher than the background composition.3 The paleoclimatic value of the snow and ice-core record in polar regions has long been recognised as an archive for the atmospheric composition in the past.4,5 Information on the historical composition of the Earth’s atmosphere is preserved in dateable polar snow, firn and ice.6,7 It has also been amply demonstrated that the degradation and long range transport of volatile organic compounds in the troposphere leads to the production of a range of secondary pollutants which may have a harmful impact on human health and on the environment8 and numerous studies have been undertaken to analyze various organic compounds in the remote atmosphere.9–11 Speciation analysis, on the other hand, has seldom been applied to such studies of the global atmospheric environment although, e.g., the speciation of organic forms of elements (Se, As, Pb, Hg…) in pristine environments can be considered a source of valuable information regarding global atmospheric transport phenomena and bio (geo) chemical cycles of the elements.12 In fact, in earlier work of our laboratory speciation of alkyllead compounds (used as antiknock additives in gasoline) measured in polar ice and snow sheets contributed to the reconstruction of the history of global pollution by automotive lead.13 This study was at the limit of analytical possibilities with detection limits at the 10 fg g21 level, controlled by the blank.14 As analytical potential in speciation analysis gradually develops, a number of other interesting topics of study can be addressed. For a description of speciation oriented analytical techniques we refer to review papers.12,15 In this paper we describe two examples of the significance of speciation analysis in remote environments and the long range transport of species in the atmosphere.First, we will outline some recent work on the analysis of organolead in snow and ice and how significant data on the long range (longitudinal) transport of these compounds can be derived.Then we will describe work on the determination of a range of naturally produced halogenated compounds and how their vertical long range transport in the stratosphere may contribute in stratospheric chemical reactions. Alkyllead compounds in recent snows from Greenland and Mont Blanc glacier The Greenland snow and ice cap has been shown to contain unrivalled archives of the large-scale atmospheric pollution of the Northern Hemisphere by lead and other elements.The first evidence of global contamination by lead was first revealed some three decades ago by Patterson and coworkers,16 who analyzed lead in snow and firn samples covering the 1753–1965 period. Their results generated considerable controversy which was resolved through a firm assessment of the natural lead levels in old Greenland ice dated 5500 years BC.17 After a 200 fold increase observed from ancient times to the late 1960s, lead levels dropped by a factor of 7.5 between the 1970s and 1992.18 This drop was tentatively explained by Boutron19 as the result of concerted initiatives in the USA and Europe to decrease the use of leaded gasoline fuel containing methylated and ethylated alkyllead compounds.In 1995, leaded gasoline still represented 34% of the European Union market; it should decrease to 5% for the year 2005 and a total phase out is scheduled for the year 2015. The organolead (OrgPb) compounds added to gasoline are apolar methylated and ethylated tetraalkylleads including sometimes mixed methyl and ethyl groups.During combustion the breakdown of the molecule is usually not complete and partly ionic trimethyllead, triethyllead (trialkyllead species), † Presented at The Third International Symposium on Speciation of Elements in Toxicology and in Environmental and Biological Sciences, Port Douglas, Australia, September 15–19, 1997. ‡ Present address: Risø National Laboratory, Plant Biology and Biogeochemistry Department PBK-124, P.O.Box 49, DK-4000, Roskilde, Denmark. Analyst, May 1998, Vol. 123 (767–772) 767dimethyllead and diethyllead (dialkyllead species) are also found in the exhaust gas. Improvements in analytical methodology for speciation analysis provided ultra-low fg g21 levels for the determination of alkyllead compounds whose presence was unambiguously proven in snow cores integrating the last decades of precipitation in Central Greenland.13 No organolead could be detected in pre-1923 samples, thereby putting an upper limit on the natural production of bio-methylated lead in the environment. This point is important: alkylated lead compounds are much more toxic than inorganic lead, a straight relation thus links the lead bioavailability and its environmental impact.A complementary side of the research focuses on the extent to which OrgPb compounds are transported towards the polar zone, and their fate during the transfer. This can be assessed by investigating the changes in the OrgPb concentrations in the successive snow layers.Lobinski et al.20 addressed this particular issue through the analysis of 18 fresh snow samples deposited between January and August 1989 at Dye 3, a Southern Greenland site, and 12 samples from Summit, Central Greenland, from May 1987, June 1989 and July 1989 snow events. In the Dye 3 samples, the total ionic OrgPb levels from January to April snow falls ranged from 100 to 800 fg g21, with an average around 470 fg g21.They were lower, from 15 to 220 fg g21 with an average of 83 fg g21, in the following May to August samples. From one snowfall to another the total concentrations varied by one order of magnitude, reflecting the differences in the OrgPb burden in the successive air masses reaching Greenland. For the Summit site, the average concentration is also rather high; if one sample is discded for which the total organolead content is much above the others, then the average reaches 460 fg g21 with a range between 250 and 900 fg g21.A peculiar feature observed at both sites is the dominance of ethylleads in the total composition of OrgPb. Methyl species were seldom observed: at Dye 3 they were found only in the winter and a few spring samples, at Summit they were observed in only one sample from mid-summer. The total OrgPb concentrations were observed to peak in winter and early spring and also to a minor extent in mid-summer, but when reported for total lead concentrations, the OrgPb:Pb2+ ratio drops from late spring to mid-summer.It indicates that alkyl species mostly transported in the vapor phase are more sensitive to react with free OH radicals and ozone, the concentrations of which increase with polar sunrise. The ratio of dialkyl- to trialkyllead peaks around mid-June and sharply decreases later; this strongly suggests a photochemical de-alkylation process. An enrichment in the OrgPb versus Pb2+ was evidenced in the Dye 3 samples, with ratios up to 2%, the higher values observed until late April then to a lesser extent in summer.In Summit samples dated in the summer, this ratio was, however, much higher and could reach up to 5 or even 10%. Lobinski et al.20 suggested that trialkyllead follows a pathway of atmospheric transport similar to SO2 and is equally persistent in the atmosphere. Vapor phase alkyllead compounds could thus play a key role in extending the presence of OrgPb in the atmosphere in the same way as SO2 does for sulfate.To provide further evidence for all this, we report here the analysis of 68 snow pit samples integrating four years of snow accumulated at the Summit site between early spring 1991 and mid-spring 1995. The snow dome is located at high altitude in the troposphere, 3200 m above sea level, which indicates that the air masses reaching its surface are an integral part of the global circulation of the Northern Hemisphere.These samples were taken along the vertical wall of a hand-dug pit by experienced operators dressed in full clean room garments including polyethylene gloves.18,21 For digging the pit and sampling, acid-cleaned polyethylene materials and sample storage containers were used in strict observance of the handling rules and protocols developed for trace metal analysis in clean rooms.21,22 Following the same procedures, 11 samples were taken from Col du D�ome, a high alpine site at 4500 m above sea level in the heart of the Mont Blanc mountains.For these samples, the first ones taken for OrgPb analysis from a high tropospheric European site, dating is less precise but to the best of our knowledge, the series extends from March or April 1994 up to mid-June 1994. These samples are valuable as here the organoleads are deposited close to the emission sources in Europe. Hence, they can provide a reference for interpreting the transformation processes the alkylleads undergo during atmospheric transport.In the Greenland samples, total OrgPb concentrations showed large seasonal changes, which reflect a combination of factors including variations in source strength, changes in atmospheric transport due to meteorology and the dealkylation process during transport (Fig. 1). Maximum concentrations are usually seen in spring with high inputs also during winter. Summer arrivals are usually cleaner, probably due to the lower stability of the compounds in the atmosphere.Compared with Dye 3 and Summit data,20 the organic fraction shows a similar level of enrichment, with the OrgPb:Pb2+ ratio below 2% for most of the samples and a few higher values up to 8% (Table 1). This is consistent with observations from remote sites such as rural Ireland and the Outer Hebrides.23,24 No clear seasonal pattern emerges from Fig. 1 but within each year, higher ratios are recorded in summer and winter than in spring.When peaks are seen in summer, they tend to occur at mid-season. Unlike the ethyllead species, methylleads are only detected in less than 40% of the samples, in which, on average, they amount to only 38% of the alkyllead content (Table 1). Methylated species are definitely minor contributors to the total organic concentration of the samples. They are present at rather low concentrations, from about 3 to 77 and 84 fg g21 for tri- and dimethyllead, respectively, and average around 20 fg g21 for both species, with higher levels in spring than in winter.Moreover, they are below the detection limit for 1992 and 1993 samples (Fig. 2). Interestingly, Lobinski et al.13 did not detect Fig. 1 Data from the Summit site, Central Greenland. a: Total organolead concentrations; b: organic Pb/inorganic Pb ratio in %. Sp: Spring, Su: Summer, W: Winter. 768 Analyst, May 1998, Vol. 123methyl species in the snow core record from Summit, but they indeed measured some, above 50 fg g21, in one particular Summit snow event from July 1989 and also in Dye 3 samples resulting from January to April 1989 precipitations.20 In our samples, ethyllead species dominate the total OrgPb concentrations.The seasonal variations of tri- and diethyllead (TEL and DEL) measured at Summit are similar to those evidenced at Dye 3 in the previous study despite the differences in geography (latitude and altitude) and meteorological conditions. 25,26 A clear trend towards higher concentrations is observed in winter and spring but on the other hand, summer layers can also record important arrivals. Lobinski et al.13 observed that for Summit, a DEL:TEL ratio below 0.5 could characterize lead species originating from the USA. This ratio exceeds 1 in most of our samples (Fig. 2) which indicates a mainly Eurasian origin. This is consistent with the virtually complete ban of lead additives in the USA. Maxima in the ratio are observed in spring; the ratio then decreases in summer due to the increasing effect of the photodegradation as explained above.To a lesser extent, maxima also occurred in winter 1992 and 1993 (Fig. 2); globally these changes in the DEL:TEL ratio with time show a more or less reproducible pattern. Considering now the Mont Blanc data, the concentrations are found to vary within a much smaller range for all the investigated species (Table 1). From the first to the last sample, Pb2+ showed a 10-fold increase.These patterns can be tentatively explained by differences in the behavior of the phase carrier. The lifting of the boundary layer in summer causes an enhancement in vertical mixing and results in more particulate material being transferred in the high troposphere. Such changes do not affect OrgPb to the same extent since they are mainly carried in the vapor phase. This interpretation is in good agreement with conclusions reached in the Alpine Alptrac project.27,28 Also, no enrichment in the organic phase is evidenced from the data and the organic/total lead ratio remains below 0.1% (Fig. 3); it indicates that the cloud of precipitation initiated in an air mass of probably regional origin. Moreover, as discussed below, the atmospheric transport is probably rather quick as no major changes in the distribution of the species seem to occur prior to deposition. The question remains to what extent the snow record reflects the emission sources.To elucidate this, samples obtained close to the source are valuable. Organoleads were measured in bottled vintage wines from a vineyard about 300 km South- West of Mont Blanc.29 This vineyard is located in the vicinity of a major North–South and East–West motorway crossing, so is likely to give a synoptic picture of the lower troposphere burden on a local scale. As the grape harvest occurs around mid- September, a slight time shift separates both records and may hamper a straightforward relation.Both records are similar in respect to the distribution between the species: methyllead species are the most abundant in both series and the methyl: ethyl ratio in the latter vintages (1987 to 1990) are very close to those observed at the Mont Blanc site (Table 1). Unfortunately, the winemaking process dilutes the atmospheric lead with foreign inputs, thus no information can be gained from the ornic : inorganic Pb ratios.30 In conclusion, the Mont Blanc and Greenland records are different in their OrgPb content despite similar altitudes.Mont Blanc probably reflects the composition of regional or eventually Western European air masses quickly transported to the high troposphere without much alteration in the original emissions. The lack of appearance of the methyl species in the Greenland samples is probably not due to specific en-route chemistry as they are more stable than the ethyl species.31 Vapor phase organoleads have a longer residence time in the atmosphere than particulate lead and the air masses content affecting Summit may represent a mixing of different sources where the Western European contribution is probably minor.Table 1 Comparison of the ratios of different organolead species obtained in this work and previously published data. When available the range and the average (in parentheses) are given Dye 3* Summit† Mont Blanc† Wine‡ MtSpc:EtSpc§ 0.03–0.57 0.02–1.48 1.49–33.3 0.5–34.9 (0.27) (0.38) (8.09) (13.7) DML:TML§ 1.8–7 0.04–3.7 0.8–2.45 — (4.4) (1.5) (1.6) DEL:TEL 0.5–20 0.2–11.6 0.21–3.13 — (5.6) (2.34) (0.83) OrgPb:InorgPb 0.01–2.02 0.03–8.7 0.02–0.33 — (%) (0.5) (1.13) (0.12) MtSPc = methyl species, total of tri- and dimethyllead; EtSpc = ethyl species, total of tri- and diethyllead; DML, TML, DEL, TEL = di- and trimethyllead, di- and triethyllead, respectively.* Lobinski et al., 1994.20 † This work. ‡Lobinski et al., 1994.29 § Only calculated when methyl species are measured. Fig. 2 Data from Summit, Central Greenland. a: Methyl species concentrations; b: diethyl Pb/triethyl Pb ratio. Sp: Spring, Su: Summer, W: Winter. Fig. 3 Data from Col du Dome, Mont Blanc area. Organolead versus. total lead, ratio in %. Sample 1: mid June 1994, sample 11: March or April 1994. Analyst, May 1998, Vol. 123 769Volatile organohalogen compounds in the global environment Since the first discovery of the formation of an ‘ozone hole’ over Antarctica by Farman et al.,32 scientists have focused on the reasons for stratospheric ozone destruction and its effects on the global environment.It has been found that photochemically formed halogen radicals were the cause of this important phenomenon of the South polar region, each year at sunrise in spring.33,34 The reason that similar high ozone decomposition has not occurred yet over the North polar region is the special meteorological condition in Antarctica.A cooling of the atmosphere in winter down to 280 °C leads to the formation of nitric acid–water aerosols (PSC = Polar Stratospheric Clouds). At the surfaces of these aerosols highly reactive halogen compounds are formed during the winter. With the appearance of the first sunlight in the Antarctic spring these ‘reservoir compounds’ are photolytically cleaved and halogen radicals are released into the stratosphere.35 These photochemically formed halogens can then decompose ozone in a catalytic process: X + O3 ? O2 + XO (1) XO + O ? X + O2 (2) O + O3 ? 2O2 (3) Chloro- and chlorofluorohydrocarbons (CFCs) released by human activities have been identified as the sources of these halogens.The widespread use of these anthropogenic halocarbons as coolants, propellants, agents for fire extinguishers, fuel additives and solvents caused a high annual input into the atmosphere. Due to the known threat to the ozone layer, several industrial nations stopped the production and use of these compounds by 1996 (the Montreal Protocol, 1989).However, as most of the anthropogenic halocarbons have a very long atmospheric lifetime, it would take a long time, perhaps until the middle of the next century, to reduce concentrations to what they were before the ozone hole developed. Besides this anthropogenic input of halocarbons into the atmosphere, a biogenic flux also exists (Fig. 4).36,37 Whereas the anthropogenic input consists mainly of chlorinated compounds, the biogenic part is dominated by brominated and iodinated hydrocarbons.Brominated compounds in particular have received considerable interest recently.38 Like the CFCs, some of the bromocarbons, e.g., methyl bromide, are stable enough to reach the lower stratosphere before they are photolytically decomposed and serve as a halogen source.39 However, compared to chlorine, bromine is about 50 times more effective in destroying stratospheric ozone and, therefore, would be a much greater threat to the ozone layer.38 Stopping the use of industrially produced halocarbons would reduce the anthropogenic flux into the atmosphere.However, as long as the biogenic flux and its part in atmospheric photochemical reactions remain unknown, estimation of the end of the stratospheric ozone destruction is hardly possible. In contrast to the input of anthropogenic volatile halocarbons, which can be calculated accurately from industrial production data, the biogenic input is difficult to estimate because their sources have not been fully explored yet.Several scientists36,37,40 reported the oceans as an important source of biogenic halocarbons (Fig. 4). Furthermore, terrestrial sources like forest soil may have to be considered as contributors to the halocarbon flux.41 In the oceans, marine macroalgae have been identified which are capable of producing and releasing volatile halocarbons.42,46 However, as macroalgae are restricted to coastal areas, they may not be solely responsible for the halocarbon concentrations detected in the open oceans.47 Recently, Tokarczyk and Moore48 reported the release of halocarbons by unialgal cultures of marine phytoplankton. Field data from the open oceans about the implications of phytoplankton producing volatile halocarbons are still missing.Thus, an extrapolation from these controlled culture experiments to the marine environment cannot be done yet. Attention has been focused on halocarbon release by macroalgae located in the polar regions.Especially in Antarctica, where stratospheric ozone depletion reaches high levels, macroalgae occur down to considerable depths ( > 30 m). They occur along thousands of kilometers of coastline in the Antarctic peninsula,49,50 and would have a much higher influence on halocarbon release into the atmosphere than algae from temperate regions. The investigation of several species of red, green and brown Antarctic macroalgae in field and culture experiments showed the production and release of a wide range of volatile halogenated compounds (Table 2).45,51 It is interesting that many halocarbons, e.g. methyl bromide (used for soil fumigation) and 1,2-dibromoethane (used as a gasoline additive), which were believed to have an anthropogenic origin only, are also formed biogenically.Among the compounds found, bromoform dominates halocarbon release.It is released in up to 20–30-fold higher quantities compared to other major compounds released like dibromomethane and dibromochloromethane (Fig. 5). The release of volatile halocarbons by polar macroalgae is dominated by brown and green algal species (Fig. 5),45,46 whereas red algal species showed only low release. As this was reported also for temperate algal species,44 red macroalgae seem to play only a minor role with regard to halocarbon input into the global environment.However, since Fig. 4 Known sources for biogenic volatile halogenated organic compounds and their contribution to global atmospheric reactions. Phytoplankton has been marked with a dotted line, as its contribution has been investigated in culture experiments only.38,39,41–46,48 Table 2 Various volatile halogenated hydrocarbons and their release rates determined for 59 different polar macroalgae. Values are the average release rates and the variation of the release rates for the single algal species45,46,51,67 Release rate/pmol g21 wet algal weight d21 Compound Formula Average Range Methyl chloride CH3Cl 35 0–2.8 3 103 Methyl bromide CH3Br 1.9 0–67 Bromochloromethane CH2BrCl 9.2 0–85 Dibromomethane CH2Br2 90 0–693 Bromodichloromethane CHBrCl2 23 0–145 Dibromochloromethane CHBr2Cl 48 0–600 Bromoform CHBr3 1.2 3 103 0–15.4 3 103 Bromoethane C2H5Br 719 0–3.6 3 103 1,2-Dibromoethane 1,2-C2H4Br2 41 0–266 Methyl iodide CH3I 1.7 0–63 Chloroiodomethane CH2ClI 3.8 0–38 Diiodomethane CH2I2 68 0–183 770 Analyst, May 1998, Vol. 123the first investigations of halocarbons formation by marine macroalgae, various assortments of halogenated organic compounds have been found in the extracts of red macroalgae.52,54 Thus, red algae were regarded as an abundant source of halocarbons. Apparently, they can synthesize a wide range of halogenated compounds, but as contributors to the input of volatile halocarbons into the global environment, they may be of less importance.Details about formation mechanisms of halocarbons and their relevance to algal metabolism are of importance. Unfortunately, their function and formation are still poorly understood. So far, the formation of halocarbons by means of an enzyme-controlled mechanism is assumed.55 Haloperoxidases, an enzyme group that has been detected in a wide range of marine and terrestrial organisms,56,59 can catalyze the oxidation of halogens in the presence of hydrogen peroxide to form halogenated organic compounds.60 Metabolic pathways by which volatile halocarbons such as bromoform are synthesized have been discussed by several authors.52,55,61 Intracellular halogenation of ketones present in marine algae followed by decay via the haloform reaction can lead to the formation of polyhalogenated methanes like bromoform and dibromomethane.61,62 Another pathway may be the reaction of hypobromous acid, an extremely reactive species, with organic matter to form volatile halocarbons.Hypobromous acid can be formed by haloperoxidases located near the macroalgal surface and then released into seawater.58 Although the sites of high halogenating activity correlate well with the release sites of high quantities of halocarbons,46,63 i.e. the formation of volatile halocarbons is clearly connected to the enzyme activity, the exact formation mechanisms of most of the volatile halogenated C1–C4 hydrocarbons remain unknown.Furthermore, nothing can be said yet about the function of volatile halocarbons in algal life. Fenical52 pointed out that they may be a chemical defense against microorganisms or herbivores, and Gibson et al.64 described the narcotic effects of bromoform on marine organisms. However, it is possible that these small molecules have no particular function, and, perhaps, are only decomposition products in algal metabolism. Several Antarctic marine macroalgae were investigated in culture and field studies for their release of 12 volatile halocarbons ranging from chloromethane to bromoform. Determination and verification were carried out by a purge-and-trap gas chromatographic method.A capillary GC instrument was connected to an electron capture detector (ECD) and a microwave induced plasma atomic emission detector (AED), respectively. Both detectors were compared for their analytical characteristics in the determination of volatile halocarbons. Whereas with the ECD very low detection limits down to 1 pg l21 were obtained, the multichannel AED has the advantage of being able to detect simultaneously two or more compounds with the same retention time and moreover, provides a considerably more specific identification.Table 3 gives a summary of the analytical characteristics. Antarctic macroalgae have been found to produce and release several volatile halogenated organic compounds and are an important biogenic source of these compounds in the environment.Due to their occurrence in the Antarctic region only and their, apparently, lower release of halocarbons compared to temperate macroalgae, Antarctic algae may not be a major contributor to the annual world-wide halocarbon production. However, the fact that they are located in an area known for high destruction of the stratospheric ozone layer makes them a considerable source for halides in the Antarctic environment. Recently, during the investigation of the dependence of halocarbon release on varying light intensities, it was found that Antarctic algae exhibit higher halocarbon release at lower light intensities and in the dark,65 i.e.at conditions occurring during the Antarctic winter.66 In addition, mostly higher rates were found for methyl halide release at short day-lengths. This may be important for the contribution of Antarctic macroalgae to the atmospheric chemistry in the South polar region. If a higher release of volatile halomethanes occurred during the winter, the contribution of naturally produced volatile halocarbons to the destruction of the ozone layer, which appeared over Antarctica in spring after sun rise,33 would be much higher than generally assumed.Conclusions The two examples given show that speciation analysis is able to provide information on long range transport mechanisms in the atmosphere which cannot be obtained from elemental analysis as parameters become available that help in the interpretation of the phenomena of interest.Both the examples given are based on the application of wet chemical techniques using hyphenated methods of analysis combining gas chromatography and spectrometric detection.12 For organoleads, seasonality in meteorological conditions and sunlight appear as important parameters modulating the atmospheric transport from source areas to the Arctic. However, the differences in the Mont Blanc and Greenland records suggest the Western European sources are only minor contributors to the global atmospheric organolead budget. Polar macroalgae are perhaps less important as contributors to the input of volatile halocarbons into the global environment.However, they may not be negligible for the atmospheric chemistry in the polar regions. Of course, a final conclusion cannot be drawn yet, as the dependence of halocarbon release by polar macroalgae on environmental Fig. 5 Release of volatile halocarbons by several species of brown, red and green Antarctic macroalgae.CHBr3 = bromoform, CH2Br2 = dibromomethane, CHBr2Cl = dibromochloromethane.45 Table 3 Sensitivity and detection limits of atomic emission detection (AED) and electron capture detection (ECD) AED ECD Absolute Relative Absolute Relative Compound LOD/pg LOD/ng l21 LOD/pg LOD/ng l21 CH2BrCl 504 50 0.30 0.009 CHBrCl2 209 20 0.22 0.007 CHBr3 120 12 2.67 0.178 CH2I2 130 13 9.34 0.778 CH3I 425 42 0.09 0.001 C2H5I 585 58 0.33 0.004 Analyst, May 1998, Vol. 123 771factors, like seasonality, temperature, nutrients composition and salinity variation due to ice melting, are still undiscovered. ‘This work was supported by the Belgian Federal Services for Programmation of Science Policy (DWTC), Brussels in the Global Change programme 1991–1995, by the FWO, Brussels through financial support to one of us (M.H.) and by the European Commission through a Marie Curie fellowship (JP. C., TMR grant ERB 4001 GT 9554423).Sampling campaigns in Greenland were financed by the French Ministry of the Environment and the US National Science Foundation. We thank also Dr. J. L. Jaffrezo from the Laboratory of Glaciology and Environmental Geophysics (LGGE), Grenoble, France for providing the snow and ice samples, and Dr. C. Wiencke from the Alfred Wegener Institute for Polar and Marine Research, Bremerhaven, Germany for the cultivation and identification of several polar macroalgae species.References 1 Heidam, N. Z., Wahlin, P., and Kemp, K., Atm. Environ., 1993, 27A, 3029. 2 Heumann, K. G., Anal. Chim. Acta, 1993, 283, 230. 3 Pacyna, J. M., Voldner E., Keeler G. J., and Evans, G., Proceedings of the 1st Workshop on Emissions and Modelling of Atmospheric Transport of Persistent Organic Pollutants and Heavy Metals, Durham, NC, 6–7 May 1993, EMEP/CCC Report 7/93, The Norwegian Institute for Air Research, Lillestrom, Norway. 4 Oeschger, H., and Langway, C.C. Jr., (eds.), The Environmental Record in Glaciers and Ice Sheets, Report of Dahlem-Berlin 1988, Wiley, Chichester, UK, 1989. 5 Delmas, R. J. (ed.), Ice Core Studies of Biogeochemical Cycles, NATO-ASI Series I, Global Environmental Change, vol. 30. 6 Hong, S., Candelone, J. P., Turetta, C., and Boutron, C. F., Earth. Planet. Sci. Lett., 1996, 143, 233. 7 GRIP Members, Nature, 1993, 364, 203. 8 Wayne, L., Chemistry of Atmosphere, Clarendon Press, Oxford, 1991. 9 Masclet, P., Cachier, H., Liousse, C., and Wortham, H., J.Atm. Chem., 1993, 12, 247. 10 Jaffrezo, J. L., and Davidson, C. I., Atm. Environ., 1993, 27A, 2703. 11 Tanzer, D., and Heumann, K. G., Int. J. Environ. Anal. Chem., 1992, 48, 17. 12 Lobinski, R., Appl. Spectrosc., 1997, 51, 260A. 13 Lobinski, R., Boutron, C. F., Candelone, J. P., Hong, S., Spuznar- Lobinska, J., and Adams, F. C., Environ. Sci. Technol., 1994, 28, 1467. 14 Lobinski, R., Boutron, C. F., Candelone, J. P., Hong, S., Spuznar- Lobinska, J., and Adams, F.C., Anal. Chem., 1993, 65, 2510. 15 Spuznar-Lobinska, J., Witte, C., Lobinski, R., and Adams, F. C., Fresenius’ J. Anal. Chem., 1995, 351, 351. 16 Murozumi, M., Chow, T. J., and Patterson, C. C., Geochim. Cosmochim. Acta, 1969, 33, 1247. 17 Ng, A., and Patterson, C. C., Geochim. Cosmochim. Acta, 1981, 45, 2109. 18 Candelone, J. P., Hong, S., Pellone, C., and Boutron, C. F., J. Geophys. Res., 1995, 100, 16605. 19 Boutron, C. F., G�orlach, U., Candelone, J.P., Bolshov, M. A., and Delmas, R. J., Nature, 1991, 353, 153. 20 Lobinski, R., Boutron, C. F., Candelone, J. P., Hong, S., Szpunar- Lobinska, J., and Adams, F. C., Environ. Sci. Technol., 1994, 28, 1459. 21 Candelone, J. P., Hong, S., and Boutron, C. F., Anal. Chim. Acta, 1994, 229, 9. 22 Boutron, C. F., Fresenius’ Z. Anal. Chem., 1990, 337, 482. 23 Allen, A. G., Radojevic, M., and Harrison, R. M., Environ. Sci. Technol., 1988, 22, 517. 24 Hewitt, C.N., de Mora, S. J., and Harrison, R. M., Mar. Chem., 1984, 15, 189. 25 Kahl, J. D. W., Martinez, D. A., Kuhns, H., Davidson C. I., Jaffrezo, J. L., and Harris, J. M., J. Geophys. Res., 1997, in the press. 26 Davidson, C. I., Jaffrezo, J. L., Small, M. J., Summers, P. W., Olson, M. P., and Borys, R. D., Atm. Environ., 1993, 27A, 2739. 27 Maupetit, F., Wagenbach, D., Weddeling, P., and Delmas, R. J., Atm. Environ., 1995, 29, 1. 28 Puxbaum, H., and Wagenbach, D., in Proceedings of the EUROTRAC Symposium ’94, ed.Borrell, M. P. et al., SPB Academic Publishing, The Hague, 1994. 29 Lobinski, R., Witte, C., Adams, F. C., Teissedre, P. L., Cabanis, J. C., and Boutron, C. F., Nature, 1994, 370, 24. 30 Rosman, K. J. R., Chisholm, W., Jimi, S., Candelone, J. P., Boutron, C. F., Teissedre, P. L., and Adams, F. C., Environ. Res., submitted. 31 Hewitt, C. N., and Harrison, R. M., Environ. Sci. Technol., 1986, 20, 797. 32 Farman, J. C., Gardiner, B. G., and Shanklin, J.D., Nature, 1985, 315, 207. 33 Solomon, S., Nature, 1990, 347, 347. 34 Anderson, J. G., Toohey, D. W., and Brune, W. H., Science, 1991, 334, 138. 35 Crutzen, J. P., and Arnold, F., Nature, 1986, 324, 651. 36 Lovelock, J. E., Nature, 1975, 256, 192. 37 Cicerone, R. J., Heidt, L. E., and Pollack, W. H., J. Geophys. Res., 1988, 93, 3745. 38 Butler, J. H., Nature, 1995, 376, 469. 39 Butler, J. H., and Rodriguez, J. M., in The Methyl Bromide Issue, ed. Bell, C. H., Price, N., and Chakrabarti, B., Wiley, Chichester, UK, 1996. 40 Singh, H. B., Salas, L. J., and Stiles, R. E., J. Geophys. Res., 1983, 88, 3684. 41 Laturnus, F., Mehrtens, G., and Grøn, C., Chemosphere, 1995, 31, 3709. 42 Gschwend, P. M., MacFarlane, J. K., and Newman, K. A., Science, 1985, 227, 1033. 43 Manley, S. L., Goodwin, K., and North, W. J., Limnol. Oceanogr., 1992, 37, 1652. 44 Nightingale, P. D., Malin, G., and Liss, P. S., Limnol. Oceanogr., 1995, 40, 680. 45 Laturnus, F., Kl�oser, H., and Wiencke, C., Mar. Environ. Res., 1996, 41, 169. 46 Laturnus, F., Mar. Chem., 1996, 55, 359. 47 Class, T., and Ballschmiter, K., J. Atm. Chem., 1988, 6, 35. 48 Tokarczyk, R., and Moore, R. M., Geophys. Res. Lett., 1994, 21, 285. 49 DeLaca, T. E., and Lipps, J. H., Antarct. Peninsula Antarct. J., 1976, 3, 12. 50 Kl�oser, H., Ferreyra, G., Schloss, I., Mercuri, G., Laturnus, F., and Curtosi, T., J, Mar. Syst., 1993, 4, 289. 51 Laturnus, F., Chemosphere, 1995, 31, 3387. 52 Fenical, W., J. Phycol., 1975, 11, 245. 53 Faulkner, D. J., in The Handbook of Environmental Chemistry–the Natural Environment and the Biogeochemical Cycles, ed. Hutzinger, O., Springer Verlag, Berlin, 1980. 54 Moore, R. E., Acc. Chem. Res., 1977, 10, 40. 55 Neidleman, S. L., and Geigert, J., Biohalogenation–Principles, Basis, Roles and Applications, Ellis Horwood Series in Organic Chemistry, Ellis Horwood, Chichester, UK, 1986. 56 Yamada, H., Itoh, N., Murakami, S., and Izumi, Y., Agric. Biol. Chem., 1985, 49, 2961. 57 de Boer, E., van Kooyk, Y., Tromp, M. G. M., Plat, H., and Wever, R., Biochim. Biophys. Acta, 1986, 869, 48. 58 Wever, R., Tromp, M. G. M., Krenn, B. E., Marjani, A., and Van Tol, M., Environ. Sci. Technol., 1991, 25, 446. 59 Harper, D. B., Biogenesis and Metabolic Role of Halomethanes in Fungi and Plants, Marcel Dekker, New York, 1993. 60 Butler, A., and Walker, J. V., Chem. Rev., 1993, 93, 1937. 61 Theiler, R., Cook, J. K., Hager, L. P., and Siuda, J. F., Science, 1978, 202, 1094. 62 Burreson, A. J., Moore, R. E., and Roller, P. P., J. Food. Chem., 1976, 24, 856. 63 Mehrtens, G., and Laturnus, F., Polar Res.,1997, 16, 19. 64 Gibson, C. I., Tone, F. C., Wilkinson, P., and Blaylock, J. W., Ozone: Sci. Eng., 1979, 1, 47. 65 Laturnus, F., Wiencke, C., and Adams, F. C., Mar. Environ. Res., 1998, in the press. 66 Kl�oser, H., Quartino, M. L., and Wiencke, C., Hydrobiologica, 1996, 333, 1. 67 Laturnus, F., Wiencke, C., and Adams, F. C., Geophys. Res. Lett., 1998, in the press. Paper 7/07185J Received October 6, 1997 Accepted December 4, 1997 772 Analyst, May 1998,

ISSN:0003-2654    

DOI:10.1039/a707185j

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Speciation of yttrium and lanthanides in natural water by inductively coupled plasma mass spectrometry after preconcentration by ultrafiltration and with a chelating resin† Hiroki Haraguchi*, Akihide Itoh, Chisen Kimata and Hajime Miwa Department of Applied Chemistry, Graduate School of Engineering, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan A combined system of size exclusion chromatography, UV absorption detection and ICP-MS was used for elucidation of the dissolved states of yttrium and lanthanide elements in lake water collected from Lake Biwa, Japan.The lake water samples were filtered with a membrane filter (pore size 0.45 mm) just after filtration with a glass filter (pore size 1.0 mm). The total concentrations of the dissolved lanthanide elements in the filtrate were determined by ICP-MS after chelating resin preconcentration. For speciation analysis, the above filtered water samples were further preconcentrated 30–500-fold with an ultrafiltration filter, which allowed preconcentration of molecules with molecular masses > 10 000.The preconcentrated sample solution was then subjected to speciation analysis using the above combined system, where a size exclusion column with the molecular permeation range between 1 000 and 300 000 was used for molecular separation. In the size exclusion chromatograms obtained with UV absorption detection at 254 nm, two peaks of some large organic molecules were obtained at retention times corresponding to molecular masses of > 300 000 (peak 1) and about 50 000–10 000 (peak 2).In the chromatograms, which were measured on-line by ICP-MS, yttrium and lanthanide elements (Y, La, Ce and Pr) were found at the two peak positions corresponding to above large organic molecules. Keywords: Speciation; lanthanide elements; natural water; inductively coupled plasma mass spectrometry; ultrafiltration In recent years, the lanthanide elements, often called rare earth elements (REEs), have been widely used in functional materials, catalysts and other products in industry, diagnosis reagents of magnetic resonance imaging (MRI) in medicine and some fertilizers in agriculture (especially in China). The amounts of lanthanide elements used have increased considerably in modern society.Consequently, the emission of lanthanide elements into the environment has also been increasing in recent years. These circumstances have resulted in an increase in our exposure to lanthanide elements and an increase in our dietary intake of lanthanide elements.Under such situations, the concentrations of lanthanide elements may be increasing even in natural waters, and hence monitoring techniques for lanthanide elements in natural waters are required for environmental protection. Furthermore, speciation analysis of trace elements is also increasingly demanded in order to elucidate their biological functions and toxicities.1,2 We have developed a method for the determination of lanthanide elements in sea-water3 and lake water4 by inductively coupled plasma mass spectrometry (ICP-MS), where a chelating resin (Chelex 100) was used for sample preconcentration.Furthermore, we have developed a method for the speciation of trace elements (mainly metal ions) in natural waters. In the speciation study, ultrafiltration was employed for the preconcentration of large organic molecules with molecular masses > 10 000.5–8 The preconcentrated samples were analysed using a coupled analytical system of size exclusion chromatography (SEC)–UV absorption detection–ICP-MS.As a result, it was found that most trace elements in natural waters exist as large organic molecule–metal complexes.7,8 In this work, the speciation method using the ultrafiltration preconcentration technique has been extended to the elucidation of the dissolved states of lanthanide elements in lake water.In addition, the total concentrations of the dissolved lanthanide elements and their vertical distribution profiles in Lake Biwa, Japan, were investigated by using a filtration technique and ICP-MS. Experimental Instrumentation A schematic diagram of the SEC–UV absorption detection and ICP-MS system is shown in Fig. 1; it is similar to the system used in previous work.7,8 The ICP-MS instrument (Model SPQ 8000A from Seiko Instrument, Tokyo, Japan) was used for the determination of trace elements after preconcentration and also for element-selective detection in SEC.The experimental conditions for the ICP-MS measurement are summarized in Table 1. In the size exclusion chromatogram measurements with ICP-MS detection, data acquisition was carried out in the multi-element mode by peak hopping over 15 m/z positions in a 5 s measurement time.7 Consequently, the chromatograms of 15 elements (or 15 m/z values) could be obtained by real-time data acquisition with the present SEC–ICP-MS system, where one data point of each chromatogram was obtained every 5 s for 2700 s (540 points).The chromatogram obtained with ICP-MS detection was smoothed by taking the average of 10 points. An HPLC instrument (Model LC-9A from Shimadzu, Kyoto, Japan) was used together with a gel filtration column (Superose 12, Pharmacia-LKB, Uppsala, Sweden). Thus, the present system was used for SEC. The molecular mass permeation range of the column used was in the range 1000–300 000.With the present SEC system, a UV absorption detector and ICP-MS were used as detectors in sequence for organic molecules and trace elements, respectively. TRIS–HNO3 buffer solution (pH 7.3) was used as the mobile phase at a flow rate of 0.5 ml min21. The sample injection volume was 200 ml. Other experimental conditions for HPLC are summarized in Table 1. Sample collection and pre-treatment The lake water samples were collected in Lake Biwa (the largest lake in Japan, located in the central part of main island).In the † Presented at The Third International Symposium on Speciation of Elements in Toxicology and in Environmental and Biological Sciences, Port Douglas, Australia, September 15–19, 1997. Analyst, May 1998, Vol. 123 (773–778) 773experiment on the vertical concentration profiles of the lanthanide elements, the water samples were collected at sampling station Ie-1, which is one of the stationary sampling sites of the Ecological Research Center of Kyoto University. The pond water samples collected from Kamiga-ike pond in the campus of Nagoya University were also used in some preliminary work.The water samples were first filtered with a glass filter (pore size 1 mm) and then with a membrane filter (pore size 0.45 mm) immediately after sampling. In the determination of the total dissolved elements, the water samples filtered with the membrane filter were acidified to pH Å 1 by adding HNO3.The acidified samples were then subjected to the chelating resin preconcentration described later. In the speciation study of the dissolved elements by SEC– ICP-MS, the water samples filtered with the membrane filter were further applied to ultrafiltration to preconcentrate large organic molecules with molecular masses > 10 000. In ultrafiltration, a ultrafilter with a molecular mass permeation limit > 10 000 was used. The preconcentrated water samples were then subjected to SEC analysis. Chelating resin preconcentration procedure The trace and ultratrace elements in lake water were preconcentrated by using a chelating resin (Chelex 100), in a similar manner to the procedure reported previously.4 The experimental procedure for chelating resin preconcentration was as follows.First, 0.3 g of the chelating resin was added to 500 ml of acidified water sample, and the pH of the sample solution was adjusted to 5 by adding acetic acid and aqueous ammonia solutions.The solution was stirred for 2 h with a magnetic stirrer. The chelating resin was collected on a glass filter and carefully rinsed with 15 ml of 1 m ammonium acetate to reduce the matrix elements (Na, K, Mg and Ca) adsorbed on the resin. Finally, the elements adsorbed on the resin were eluted with 10 ml of 2 m nitric acid. Consequently, a 50-fold preconcentration in volume was achieved with this chelating resin preconcentration. The recoveries of the elements in the chelating resin preconcentration were determined by employing the same experimental procedure for an artificial solution that contained 10 or 100 ng ml21 of trace elements, and are summarized in Table 2.These values were obtained as the means of three replicate experiments, and their precision (relative standard deviation) was within 5% for most elements. Results and discussion Total concentrations of dissolved elements in surface water of Lake Biwa The total concentrations of dissolved elements in surface water of Lake Biwa were determined by ICP-AES and ICP-MS with and without chelating resin preconcentration. The results for the surface water samples collected on different days (November 26, 1993, and August 9, 1994) are summarized in Table 2.In the present experiment, the collected surface water samples were filtered with membrane filters (pore size 0.45 mm) and acidified, as described earlier. Then Ca, Na, Mg, K, Si and Sr in the filtrate were determined directly by ICP-AES and Ba, Rb and Sb were determined by ICP-MS after 20-fold dilution of the filtrate. Other elements shown in Table 2 were determined by ICP-MS after chelating resin preconcentration, where an internal standard correction for matrix effects was made using Ge, In, Re and Tl (10 ng ml21 of each added) as the internal standard elements.As a result, analytical data for 39 elements in the lake water samples were obtained in the concentration range from 10.9 mg l21 for Ca to 0.18 ng l21 for Eu in the surface water.It is noted that all the data except Ti for the different water samples were almost identical with each other, as can be seen in Table 2. The inconsistent data for Ti in Table 2 may be due to its hydrolysis during sample storage and/or pH adjustment. The concentrations of dissolved lanthanide elements were at the low-ppt (10212 g ml21) or sub-ppt level. Since the concentrations of lanthanide elements were extremely low, it was difficult to determine them without preconcentration with a chelating resin.In both cases, Ce was present at the highest concentration and Eu at the lowest concentration among the lanthanide elements. It should be also noted that the concentrations of lanthanide elements showed a zig-zag concentration change according to the Oddo–Harkins rule.9 Vertical profiles of total concentrations of dissolved lanthanide elements in Lake Biwa The total concentrations of dissolved lanthanide elements in the lake water samples were further determined at different depths of the water column at sampling site Ie-1 in Lake Biwa. The determination of the total concentrations of dissolved lanthanide elements was performed by ICP-MS after preconcentration with a chelating resin.The water depth at sampling site Ie-1 was about 75 m, and the water samples were collected at 11 depths down to 73 m. Since the thermocline layer occurred between 25 and 35 m at the sampling time on November 26, 1993, the water Fig. 1 Schematic diagram of the LC–UV absorption detection–ICP-MS measurement system. An SEC column was used in LC for molecular size separation of large organic molecules in the natural water samples. Other instrumental components and experimental conditions are described in Table 1. 774 Analyst, May 1998, Vol. 123samples were collected at 5 m intervals between 20 and 40 m. The vertical concentration profiles of dissolved lanthanide elements (La, Ce, Pr, Yb and Lu) are shown in Fig. 2, along with that of Si. As can be seen, Si showed a characteristic vertical profile, reflecting the biological activities in the lake environment. That is, the concentration of dissolved Si was much lower above the thermocline layer (i.e., surface layer, 0–20 m), and it increased significantly in the thermocline layer and below it. It is known that Si is a nutrient element for microorganisms such as green algae and diatoms in natural water.Hence the low concentration of Si in the surface layer suggests high biological activity near the surface in Lake Biwa. On the other hand, the lanthanide elements showed the significantly different concentration profiles to Si. Furthermore, different profiles were also observed among the lanthanide elements themselves. The light lanthanide elements (La, Ce, Pr) in the dissolved form provided lower concentrations in the middle part of the water column examined, whereas Yb and Lu (heavy lanthanide elements) showed almost constant concentrations through the water column.These results suggest that the light lanthanide elements are apt to be adsorbed on the suspended particulates in water, in a similar manner to Fe and Al.8 Since the heavy lanthanide elements seem to have greater complexing abilities with the dissolved organic matter in water, as mentioned later, they may exist more in organometallic complex forms compared with the light lanthanide elements.Concentration distribution patterns of lanthanide elements in Lake Biwa The three-dimensional concentration distribution patterns of lanthanide elements are shown in Fig. 3, where the x-, y- and zaxes indicate the lanthanide elements, water depth and the normalized concentrations of lanthanide elements at each water depth, respectively. In Fig. 3, the concentrations of dissolved lanthanide elements in lake water were normalized to those in sediment collected in Lake Biwa.10 The normalized concentration distribution patterns shown in Fig. 3 are often called ‘lanthanide (or rare earth element) distribution patterns.’11,12 Since lanthanide elements show zig-zag concentration changes, as can be seen in Table 2, the normalization of their concentrations to those of chondrite (stony meteorite) or some other suitable samples is often performed to characterize the geochemical samples from their concentration distributions.In Table 1 Instrumental components and experimental conditions HPLC— Column Superose-12 (Pharmacia-LKB) Mobile phase 50 mm TRIS–HNO3 (pH 7.3) Flow rate 0.5 ml min21 Sample volume 200 ml UV absorbance measurement 254 nm ICP-MS— Instrument SPQ-800A (Seiko Instrument) Rf power 1.3 kW Carrier gas flow rate 1.0 l min21 Auxiliary gas flow rate 1.0 l min21 Coolant gas flow rate 16 l min21 Nebulizer Concentric type Sampling depth 10 mm above load coil Data acquisition (multi-element mode by peak hopping) in chromatogram measurement Measurement positions 15 m/z per measurement Dwell time 100 ms at each m/z Data plot 1 point per 5 s measurements Measurement time 2700 s (540 points) ICP-AES— Instrument Plasma AtomComp MkII (Jarrell Ash) Rf power 1.0 kW Carrier gas flow rate 0.48 l min21 Auxiliary gas flow rate 1.0 l min21 Coolant gas flow rate 20 l min21 Nebulizer Cross-flow type Observation height 18 mm above load coil Table 2 Concentrations of dissolved elements in surface water on Lake Biwa sampling site Ie-1 Concentration†/mg l21 Concentration†/ng l21 Recovery Recovery Element* Nov. 26, 1993 Aug. 9, 1994 (%) Element* Nov. 26, 1993 Aug. 9, 1994 (%) Caa 10.8 ± 0.1 10.9 mg/l — Pb 70 ± 14 46 ng/l 84.3 Naa 6.8 ± 0.2 7.2 — Sn 60 ± 7 49 32.9 Mga 2.05 ± 0.01 2.09 — Ti 50 ± 12 0.2 34.5 Ka 1.53 ± 0.04 1.6 — U 21 ± 2 27 78.2 Sia 0.261 ± 0.003 0.38 — Y 13 ± 1 3.8 79.7 Concentration/mg l21 Co 7.0 ± 0.6 6.4 86.1 Cd 4.9 ± 0.5 7.2 90.1 Nov. 26, 1993 Aug. 9, 1993 La 4 ± 1 2 79.1 Sra 41.8 ± 0.5 41 mg/l — Ce 6.5 ± 0.4 3.1 76.3 Baa 8.27 ± 0.06 7.7 — Pr 0.91 ± 0.06 0.4 78.2 Fe 4.0 ± 0.4 4 50.6 Nd 3.7 ± 0.2 2 78.2 Al 3.1 ± 0.1 7 85.8 Sm 0.9 ± 0.1 0.43 78.6 Zn 2.0 ± 0.4 0.70 93.5 Eu 0.18 ± 0.01 0.1 77.7 Rbb 1.47 ± 0.01 1.45 — Gd 1.1 ± 0.1 0.60 78.2 Mn 0.73 ± 0.05 0.41 47.2 Tb 0.20 ± 0.01 0.1 77.8 Cu 0.72 ± 0.05 0.87 92.2 Dy 1.6 ± 0.1 0.58 76.5 Sbb 0.50 ± 0.03 0.46 — Ho 0.53 ± 0.03 0.2 77.2 Mo 0.36 ± 0.02 0.35 56.8 Er 2.3 ± 0.2 1.0 77.5 Ni 0.30 ± 0.92 0.92 92.4 Tm 0.45 ± 0.03 0.3 77.0 V 0.2 ± 0.01 0.2 70.2 Yb 4.1 ± 0.5 2.8 78.2 Lu 0.79 ± 0.05 0.65 78.0 * Elements with superscript a were determined by ICP-AES without preconcentration and those with superscript b were determined in the 20-fold diluted sample solutions.Other elements were determined by ICP-MS after chelating resin preconcentration. † Means of the observed values in quadruplicate reported measurements. Water samples were collected twice, on November 26, 1993, and August 9, 1994.Analyst, May 1998, Vol. 123 775the present experiment, such normalization was made with the lake sediment. This helps to find some correlation between the concentrations of lanthanide elements in lake water and sediment, especially in relation to the elemental partitionings in the aquatic environment. For the lanthanide distribution patterns in Fig. 3, some tendencies of the distributions of lanthanide elements between lake water and sediment can be discussed from the differences in the normalized concentrations.As can be seen in Fig. 3, in general, the heavy lanthanide elements provided the higher normalized concentrations. These results indicate that the heavy lanthanide elements are relatively more partitioned in water than the light lanthanide elements. These facts suggest that the heavy lanthanide elements can be more dissolved in water than the light lanthanide elements. In another words, the heavy lanthanide elements have greater complexing abilities with organic matter in water.It is also noted that anomalies with Ce are found in the distribution patterns shown in Fig. 3. That is, Ce shows relatively lower concentrations compared with the neighboring or adjacent elements (La and Pr). Such concentration anomalies are often observed in natural waters such as sea-water,3 and are generally caused by the differences in their oxidation states e.g., +4 for Ce and +3 for other lanthanide elements.13 Preconcentration of large organic molecules by ultrafiltration As preliminary work on speciation, first the preconcentration of large organic molecules with molecular masses > 10 000 were examined by using the ultrafiltration filter.In Fig. 4, the ICP mass spectra in the m/z range 120–145 are shown for cases with and without ultrafiltration preconcentration for pond water samples collected from the Kagamiga-ike pond. The peaks (ion counts) of La, Ce and Pr became larger in accordance with the 30- and 220-fold preconcentration, whereas the peaks of Sb did not change on preconcentration by ultrafiltration. These results suggest that lanthanide elements in natural waters may exist as metal complexes combined with large organic molecules.For Sb, it is considered that main species may be a small molecule of antimonate ion in a soluble form. Further investigation was therefore carried out on lanthanide elements. Distributions of lanthanide elements in large and small organic molecules As mentioned previously, lanthanide elements were enriched in the large organic molecule (M!10 000) fraction when the pond water samples were preconcentrated by ultrafiltration.Therefore, the extents (%) of the lanthanide elements, including other elements, in the large organic molecule fraction were determined in terms of the surface water samples from Kagamiga-ike pond. In this experiment, the concentrations of the elements in the preconcentrated solution (large organic molecule fraction) and the filtrate solution (small organic molecule fraction) after ultrafiltration were determined by ICP-MS, and the percentages of the elements in the preconcentrated solution and filtrate solutions were calculated from their determined concentrations.The distributions of diverse elements and lanthanide elements (including U) in the large and small organic molecule fractions are summarized in Fig. 5(a) and (b), respectively. As can be seen, the total contents of some elements (Cu and some lanthanide elements) were < 90%. However, the data in Fig. 5 Fig. 2 Vertical profiles of total concentrations of dissolved Si and lanthanide elements. Fig. 3 Three-dimensional distribution patterns of lanthanide elements in Lake Biwa. The concentrations of lanthanide elements in lake water were normalized to those in lake sediment collected from Lake Biwa, and they are expressed as ‘dissolved/sediment’ in the figure. 776 Analyst, May 1998, Vol. 123are of interest for evaluating the dissolved states of the elements in natural waters. More than 90% of the alkali and alkaline earth elements are in the small molecule fraction. This indicates that almost all of these elements exist as the simple ionic forms in natural water. Molybdenum and U also provided the large percentages of small molecules. This result can be understood from the fact that Mo exists mostly as the oxo-anion MoO4 22.For U, the existence of an oxo-cation such as UO2 2+ is suggested. In terms of the large organic molecule fractions, further speciation work was carried out by using the SEC–UV absorption detection– ICP-MS system. SEC separation of large organic molecules combined with lanthanide elements In this experiment, water samples from Lake Biwa was used for analysis. The lake water samples preconcentrated by ultrafiltration were further analyzed using the SEC–UV absorption detection–ICP-MS system.A 200 ml volume of the 500-fold preconcentrated water sample was injected into the SEC column and the UV absorption at 254 nm and the ICP-MS ion counts were sequentially measured to obtain the chromatograms for organic molecules and metallic elements, respectively. In Fig. 6, some typical chromatograms of the elements examined are shown together with the UV-detected chromatogram. Since the concentrations of lanthanide elements in lake water were extremely low, clear chromatograms were obtained only for Y, La, Ce and Pr.As can be seen in Fig. 6, two peaks of Y–Pr were commonly observed at retention times of about 700 and 1600 s. As reported previously,4,7 the two common peaks in Fig. 6 correspond to the large organic molecules observed in UV absorption detection, and their molecular masses were > 300 000 and 50 000–10 000, calculated from the molecular Fig. 4 ICP mass spectra of the pond water samples with and without ultrafiltration preconcentration. Samples collected from Kagamiga-ike pond were used.No preconcentration; (b) 30-fold preconcentration; (c) 220-fold preconcentration. Fig. 5 Distributions of trace elements in large and small organic molecules in lake water. The large and small organic molecules indicate the molecules which exist in the preconcentrated water and the filtrate water after ultrafiltration, where an ultrafiltration filter with a filtration limit of 10 000 was used.Samples collected from Kagamiga-ike pond were used. (a) Diverse elements; and (b) lanthanide elements. Black bars, Large organic molecule fractions; dotted bars, small organic molecule fraction. Fig. 6 Size exclusion chromatograms of Fe, Zn, Y, La, Ce and Pr obtained with UV absorption detection and ICP-MS. Sample: 100-fold ultrafiltration- preconcentrated lake water from Lake Biwa. Analyst, May 1998, Vol. 123 777mass calibration curve obtained by using several proteins with different molecular masses.These large organic molecules are possibly humic substances in lake water. This was demonstrated experimentally in previous work,6 in which commercially available humic substances purchased from Tokyo Kasei. (Tokyo, Japan) provided chromatograms similar to those in Fig. 6. As can be seen in Fig. 6, Fe provides only one peak at a retention time of about 700 s (Mr > 300 000), and Zn gives one peak at a retention time of about 1600 s (Mr 10 000–50 000).Further, Y, La, Ce and Pr provide two peaks corresponding to those of Fe and Zn, respectively. According to the previous work,8 the peak near 700 s is ascribed to colloidal particulate matter consisting of iron and aluminum hydroxides, maybe adsorbed by large organic molecules. On the other hand, the peak near 1600 s may originate from protein-like organic molecules, which can form metal complexes with Zn, although the chemical structures of such organic molecules have not been elucidated.In the present experiment, chromatograms for lanthanide elements except La, Ce and Pr in lake water could not be obtained with ICP-MS detection because of their extremely low concentrations. However, it can be speculated from the results in Fig. 5 that other lanthanides may exist in dissolved forms similar to La, Ce and Pr. These results indicate that the lanthanide elements exist in lake water mainly as two chemical forms binding with the colloidal particulate matter and protein-like organic molecules, which provide molecular masses > 10 000 in SEC.Conclusion The dissolved states of yttrium and lanthanide elements in natural water samples (Lake Biwa, Japan) were investigated by speciation analysis. The total concentrations of dissolved lanthanide elements, their vertical concentration profiles and normalized concentration distribution patterns and the distributions in large and small organic molecule–metal complexes were examined by ICP-MS using several sample pre-treatment and filtration techniques.From the results, several important conclusions were drawn in terms of the dissolved states of lanthanide elements in natural water. First, among lanthanide elements (present at the low-ppt or sub-ppt level), light lanthanide elements are relatively more adsorbed on or form particulate matter in lake water, whereas heavy lanthanide elements form organo-metallic complexes.These chemical properties of lanthanide elements are reflected in their vertical concentration profiles and the normalized concentration distribution pattern. Second, more than 50% of lanthanide elements are in large organometallic complexes with molecular masses > 10 000, and they can be preconcentrated by ultrafiltration. These experimental results for lanthanide elements are helpful in understanding or elucidating their chemical characteristics, kinetic behaviors, elemental cycles, etc., in the aquatic environment.It is further necessary to explore analytical methods for elucidating the chemical forms and structures of large organometallic complexes at extremely low concentration levels, which may be closely related to humic substances in natural waters. This research was supported by a Grant-in-Aid for Scientific Research (No. 08308034) (H.H.) from the Ministry of Education, Science, Sports and Culture, Japan, and also by a Sasagawa Scientific Grant (A.T.). The authors express their sincere thanks to Professor Eitaro Wada and his colleagues at Kyoto University for their kind help with water sampling in Lake Biwa. References 1 Vela, N. P., Olson, L. K., and Caruso, J. A., Anal. Chem., 1993, 65, 585A. 2 Vandecasteele, C., and Block, C. B., Modern Methods for Trace Element Determination, Wiley, Chichester, 1993. 3 Sawatari, H., Toda, T., Saizuka, T., Kimata, C., Itoh, A., and Haraguchi, H., Bull. Chem. Soc. Jpn., 1995, 68, 3065. 4 Itoh, A., Kimata, C., Miwa, H., Sawatari, H., and Haraguchi, H., Bull. Chem. Soc. Jpn., 1966, 69, 3469. 5 Itoh, A., Aikawa, M., Sawatari, H., Hirose, A., and Haraguchi, H., Chem. Lett., 1993, 1017. 6 Itoh, A., and Haraguchi, H., Chem. Lett., 1994, 1627. 7 Haraguchi, H., Itoh, A., and Kimata, C., Anal. Sci. Technol., 1995, 8, 405. 8 Itoh, A., and Haraguchi, H., Anal. Sci., 1997, 13, Suppl., 379. 9 Lee, J. D., Concise Inorganic Chemistry, Chapman and Hall, London, 1991. 10 Toyoda, K., Haraguchi, H., and Fuwa, K., Int. Project Paleolimnol. Late Cenozoic Climate Newsl., 1989, 5, 24. 11 Toyoda, K., and Haraguchi, H., Bull. Chem. Soc. Jpn., 1987, 60, 933. 12 Masuda, A., Geochem. J., 1975, 9, 183. 13 Byrne, R. H. B., and Binger, L. S., Geochim. Cosmochim. Acta, 1989, 53, 1475. Paper 7/08253C Received November 17, 1997 Accepted February 19, 1998 778 Analyst, May 1998, Vol. 123

ISSN:0003-2654    

DOI:10.1039/a708253c

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Determination of methylmercury in two mussel tissue Standard Reference Materials by pre-irradiation separation and neutron activation analysis† Elizabeth A. Mackey* and Donald. A. Becker Analytical Chemistry Division, National Institute of Standards and Technology, Building 235, B117, Gaithersburg, MD 20899, USA A method was developed for the measurement of organically bound mercury in two mussel tissue Standard Reference Materials (SRM): SRM 2974 Organics in Mussel Tissue and SRM 2976 Mussel Tissue.Organomercury compounds were distilled from a mixture of powdered sample (approximately 200 mg), sulfuric acid and copper(ii) sulfate, and bound chemically to cysteine fixed on to filter-paper. The filter papers were subjected to neutron activation analysis (NAA) for mercury determination. Other analytical methods were used to confirm that most (499 %) of the organically bound mercury is present in the form of methylmercury for these two mussel reference materials.With this established, the results from the determination of organomercury compounds for this work are reported as the amount of mercury in the form of methylmercury. The average and standard deviation of the results of analyses of four aliquots of Reference Material IAEA- 350 Tuna were 3.61 ± 0.20 mg g21 [with a 95% confidence interval (95% CI) of ±0.69 mg g21], compared with a certified value of 3.65 mg g21 and 95% CI of 3.32–4.01 mg g21. The concentration average and standard deviation from analyses of seven aliquots of SRM 2976 Mussel Tissue were 27.9 ± 3.9 ng g21 (95% CI, ± 5.2 ng g21); results from analyses of five aliquots of SRM 2974 yielded 78.3 ± 9.6 ng g21 (95% CI, ± 15.0 ng g21).Results obtained using this method compared well with results obtained by other investigators using three different analytical methods, and were used in the certification of methylmercury concentrations in these two SRMs. Keywords: Conway dish acid distillation; methylmercury; mussel tissue; neutron activation analysis; Standard Reference Material The importance of the chemical form or species of a trace element in defining its toxicity to various organisms, its fate in the environment, biological pathways and many other parameters is well known.For some trace elements, the highest oxidation state is the most toxic form. For example, chromium( vi) compounds are toxic whereas Criii is essential for humans. In contrast, arsenic(iii) is more toxic than the naturally occurring arsenic(v).For arsenic, organically bound forms such as arsenobetaine are much less toxic than the elemental forms. However, for mercury, the organically bound form of methylmercury is more toxic than the free metal. Organic forms of mercury are better able to pass through the blood–brain barrier1 and through the human placenta2 and toxicological effects associated with exposure to this compound are well documented (e.g., see ref. 3). Therefore, methylmercury is of great interest owing to its known toxicity and to its prevalence in the marine environment. Although the knowledge that any detrimental effects of a given element may vary drastically depending upon the chemical form of the element has existed for decades, analytical measurements of many of the relevant chemical forms are far from routine. In part, this may be attributed to the inherent difficulty of isolating the analyte without changing it or its oxidation state in the process.For example, although methylmercury is released from most matrices by acid leaching, care must be taken in the choice and strength of the acid. Horvat et al.4 reported the decomposition of methylmercury by hydrochloric acid of concentrations 44 mol l21 within a few days. In studies designed to determine the stability of methylmercury in distilled water and acidified distilled water, Leermakers et al.5 found that MeHg is stable for at least 2 weeks in HNO3- acidified distilled water.Stoeppler and Mathes6 found considerable degradation of the compound in HNO3-acidified seawater. In general, the stability of methylmercury in water depends upon pH, temperature and exposure to light. The stability of methylmercury in environmental samples also varies depending on the matrix. Methylmercury is sensitive to ultraviolet light and is somewhat volatile, so that over time the concentration of methylmercury may change depending on the storage conditions. For example, as mentioned above, the compound is not stable in seawater and appreciable photodegradation has been documented in surface lake waters.7 Horvat and Byrne8 reported that for some matrices repeated freezing and thawing can result in loss of methylmercury.In the same work, they reported that long-term storage of bivalve tissues at 220 °C results in methylmercury losses of about 30%, but that fresh and dried fish muscle tissue is stable under the same storage conditions.There are bacteria capable of methlyating inorganic mercury and others capable of demethylation of methylmercury so that sterilization of the material is required to ensure stability of methylmercury in many environmental matrices. Emteborg et al.9 reported that freeze-dried, sterilized sediments are very stable with regard to methylmercury content. Similar results were obtained by Horvat et al.,4 who reported that wet sediments were stable for at least eight months if stored at either +4 or 220 °C.In the same study, over the course of several years, Horvat et al. also verified the stability of four certified reference materials: National Research Council Canada (NRCC) DORM-1 Dogfish Muscle, TORT-1 Lobster Hepatopancreas and DOLT-1 Dogfish Liver and International Atomic Energy Agency (IAEA) MA-B- 3/TM Fish Tissue Lyophilized. The use of certified reference materials is an indispensable part of analytical method development.Lack of such materials can greatly hinder the progress of method development, especially in the determination of an analyte that is not particularly stable when separated from the matrix. Beyond the method development stage, certified reference materials are necessary to provide quality assurance for routine measurements. There are increasingly more reference materials for † Presented at the Third International Symposium on Speciation of Elements in Toxicology and in Environmental and Biological Sciences, Port Douglas, Australia, September 15–19, 1997.Analyst, May 1998, Vol. 123 (779–783) 779which certified values for methylmercury are provided. Currently, there are a total of 14 reference materials for which methylmercury concentrations are certified and one material for which there is a non-certified information value provided (Table 1). In general, the certification process at NIST requires the use of data from a minimum of two independent analytical techniques. The objective of this work was to develop a method for the determination of organic mercury in the two NIST mussel tissue SRMs that differed from the other method used at NIST in which organic mercury compounds are extracted into toluene from the acid leachate of a sample, separated by gas chromatography (GC), and quantified using atomic emission detection (AED) of mercury.10 In the method reported here, both the separation procedure and the detection method had no common sources of error with those associated with the GC– AED technique. Based on the results from the GC–AED method used at NIST, the results from two additional independent laboratories, and this work, NIST recently provided certified concentrations for methylmercury in two standard reference materials (SRMs), SRM 2974 Organics in Freeze-Dried Mussel Tissue and SRM 2976 Mussel Tissue.11 The first step in the determination of methylmercury in natural matrices involves the separation or isolation of the methylmercury from the matrix.Many of the methods currently used for the isolation of methylmercury fall into one of three categories: those using acidification of the sample followed by solvent extraction;10,12–14; those using acidification followed by ion exchange separation;15 and those using acid distillation followed by solvent extraction.4,16 –21 For the analysis of hair, alkaline digestion prior to acidification followed by solvent extraction has been employed to increase yields.22,23 The separation method that was used for this work is based on acid distillation. This approach was originally developed by West�o�o16 during the late 1960s, modified by Zelenko and coworkers. 17,18 in the 1970s and further developed during the past decade by Horvat and co-workers4, 19 2 21. In these most recent methods, methylmercury is distilled from the sample, bound to cysteine, subsequently released from the cysteine and extracted into either benzene or toluene in preparation for GC (with or without ethylation).The methylmercury is determined by measuring mercury using either atomic fluorescence spectrometry (AFS) or cold vapor atomic absorption spectrometry (CVAAS). In this work, the method developed involves pre-irradiation separation of organic mercury compounds from the matrix followed by determination of mercury in the separated fraction by neutron activation analysis (NAA).The organic mercury compounds are distilled from the sample and bound chemically to cysteine that is fixed on to filter-paper. The filter paper is then subjected to NAA for mercury determination. Analyses of reference material IAEA-350 Tuna were included with the analyses of SRMs 2974 and 2974 for the purpose of quality control. For many materials, most of the organomercury compounds are present in the form of methylmercury. However, using the method described in this paper, one cannot distinguish the forms of organic mercury that may be present on the filter-papers.For SRM 2976 Mussel Tissue, SRM 2974 Organics in Freeze-dried Mussel Tissue and IAEA-350 Tuna, results from other techniques used at NIST and elsewhere have demonstrated that essentially all of the organically bound mercury is in the form of methylmercury.24 Therefore, results presented here for these materials are reported as the amount of mercury that is from methylmercury.Results from other laboratories are also shown for comparison. Experimental Distillation Method An aliquot of powdered material weighing 200–250 mg was placed in the interior of a Conway dish together with 0.5 ml of 8 mol l21 ultra-pure sulfuric acid and 0.5 ml 8 mol l21 ultra-pure sulfuric acid saturated with copper(ii) sulfate. A schematic representation of a Conway dish is shown in Fig. 1. Two 1 3 5 cm strips of filter-paper, each containing 1 mg of cysteine and 20 mg sodium citrate, were placed in the outer area of the Conway dish.The outer rim of the Conway dish was moistened with distilled water to form the initial seal of the ground-glass rim to the lid. The covered dish was then placed on a hot-plate at a surface temperature of 80 °C for 6 h. The filter-papers were then removed and 50 ml of ultra-pure water were added to each to soften the paper to facilitate placing each paper into an acidwashed quartz vial (vial diameter 0.5 cm and length 7 cm).The vials were flash frozen in liquid nitrogen prior to sealing to avoid evaporative loss of volatile mercury compounds. A solution containing a known amount of mercury was pipetted into each of several quartz vials each containing a similarly sized piece of clean filter paper; these were flash-frozen and sealed for use as standards. The standards were prepared in a separate laboratory to avoid the potential for cross-contamination of the samples with inorganic mercury.Table 1 List of standard reference materials with certified values for methylmercury concentrations Organization Reference material Certified value/ mg g21 as Hg Community Bureau of Reference (BCR), Belgium CRM-422 Cod Muscle CRM-463 Total and Methylmercury in Tuna Fish CRM-464 Total and Methylmercury in Tuna FIish 0.43* 3.04 ± 0.15 5.5 ± 0.2 National Research Council of Canada (NRCC) DOLT-2 Dogfish Liver Tissue DORM-2 Dogfish Muscle LUTS-1 Non-defatted Lobster Hepatopancreas TORT-2 Partially-defatted Lobster Hepatopancreas 0.693 ± 0.053 4.47 ± 0.32 0.063 ± 0.004 0.152 ± 0.013 International Atomic Energy Agency (IAEA) IAEA-350 Tuna Homogenate IAEA-356 Polluted Marine Sediment IAEA-085 Mercury, Methylmercury and Trace Elements in Spiked Hair IAEA-086 Mercury, Methylmercury and Trace Elements in Unspiked Hair IAEA-142, 6 Mercury, Methylmercury and Organic Microconstituents in Mussel Homogenate 3.65 (3.32–4.01) 0.00546 ± 0.00038 22.9 (21.5–24.3) 0.258 (0.226–0.290) 0.047 (0.043–0.051) National Institute for Environmental Studies (NIES), Japan CRM No. 13 Human Hair 3.8 ± 0.4 National Institute of Standards and Technology (NIST), USA SRM 2974 Organics in Freeze-Dried Mussel Tissue SRM 2976 Mussel Tissue 0.0772 ± 0.0038 0.0277 ± 0.0020 * Non-certified valve. 780 Analyst, May 1998, Vol. 123Method for neutron activation analysis Each sealed quartz vial containing either one cysteine-loaded filter paper from a sample distillation, an elemental mercury standard or control material was sealed in plastic tubing to provide cushioning of the vial and to contain the sample in the event that the quartz vial broke during irradiation or transport to and from the reactor.The plastic-encapsulated vials were placed in a polyethylene irradiation vessel. Each vessel contained up to eight vials. The vessels were irradiated individually for either 1.0 h at a neutron fluence rate of 1.0 3 1014 cm22 s21 or for 2.5 h at 3.531013 cm22 s21.The samples required at least a 14 d decay time to eliminate the high count-rate from 24Na (from the sodium citrate buffer). Gamma radiation was collected using a germanium detector and associated electronics for a minimum of 24 h from each sample and for 0.5–4 h from the standards and control materials. Data were processed using a mVAX 3400 computer with Nuclear Data software. Quantification was based on comparison with the standards. Results and discussion Initial experiments were conducted on three bivalve reference materials (SRM 1566a Oyster Tissue, SRM 2974 Organics in Freeze Dried Mussel Tissue and SRM 2976 Mussel Tissue) and one soil reference material (SRM 2710 Montana Soil).The results showed that the acid distillation released all of the organic mercury from the matrix, but the degree to which inorganic mercury was also released and trapped on the cysteine paper varied depending on the matrix. For the soil material, a small percentage (0.7%) of inorganic mercury distilled over with the organic mercury compound(s) so that additional cleanup steps were required.For the mussel tissue reference materials, additional steps were not required (see detailed discussion below). For SRM 2710 Montana Soil, most of the 32.6 mg g21 of total mercury is probably in the form of inorganic mercury as is generally the case for soils.25 For the mussel tissue reference materials, 44–45% of the mercury is in the form of methylmercury.These mussel tissue reference materials also contain considerably less total mercury. The total mercury concentrations for these materials are 61 ng g21 for SRM 2976, and 176 ng g21 for SRM 2974. Both of these factors may affect the percentage of inorganic mercury that distills out of the sample under these conditions. Cysteine papers blanks and cysteine papers from sample-free (reagent only) distillations were also analyzed. It was at this point in the method development that a problem with the reagents was discovered.Measurements of two cysteine-loaded filter-papers and two papers from one reagent blank distillation showed that each cysteine-loaded filter-paper was contaminated with 9 ± 2 ng g21 of mercury. After subtracting 18 ng g21 (two filter-papers per distillation) from the total amount of mercury present on the cysteine-loaded filter papers to account for the mercury contamination, the values for the bivalve materials compared well with those reported by the other NIST group and by other laboratories (Table 2).These initial results indicated that additional separations were not necessary for the determination of organic mercury in bivalve tissues. After extensive cleaning of the laboratory hand glassware, and preparation of new reagents and new cysteine-loaded filter papers, experiments were resumed. Analysis of cysteine paper blanks showed no measurable contamination (55 ng g21) and this new batch was used for all subsequent work.One aliquot of SRM 2976 was included in the analyses with the distillation blanks and cysteine papers. The result for this one aliquot, 27.4 ± 2.0 ng g21, agreed well with the results reported by the other laboratories. Additional aliquots of all of the materials (SRMs 2974 and 2976; IAEA-350 Tuna) were subjected to this distillation and NAA procedure to determine the degree of reproducibility of the method. The results indicated that the method was reproducible and the values obtained were in agreement with those reported by other laboratories.Aliquots of IAEA-350 Tuna were included in the analysis batches to provide quality assurance for the entire newly developed method of pre-irradiation separation of methylmercury followed by NAA. Instrumental NAA of powdered aliquots of SRM 2710 Montana Soil were included in all analysis batches for the purpose of quality assurance in the quantification of mercury.Results from analyses of IAEA-350 and SRM 2710 control materials are given in Table 3. The results are in agreement with the certified values, within the uncertainties shown. The limit of detection as defined by Curie26 for this procedure ranges from 5 to 10 ng g21, depending on the magnitude of the background count-rate. Fig. 1 Schematic representation of a Conway dish. Table 2 Results of preliminary measurements using Conway dish distillation. Concentrations are expressed as ng Hg g21 dry mass, corrected for a blank of 18 ng of Hg for two filter-papers Reference This work/ From Donais and co-workers10,24 material ng g21 Values/ng g21 Laboratory SRM 1566a 19 ± 3 16.8 ± 0.5 Research Centre of J�ulich 15 ± 4 15.3 ± 4.3 NIST Not certified — SRM 2974 72 ± 5 80 ± 3 IAEA 75 ± 4 71.7 ± 1.5 Research Centre of J�ulich 80.4 ± 8.4 NIST SRM 2976 29 ± 5 28.1 ± 1.5 Research Centre of J�ulich 26.8 ± 1.5 IAEA 28.3 ± 2.2 NIST Table 3 Results of analyses of reference materials Reference material Method and analyte Measured values/ mg g21 Certified value/ mg g21 SRM 2710 INAA for Hg (mg g21 dry mass) 31.1 ± 0.3 31.0 ± 0.3 32.0 ± 0.2 30.5 ± 0.3 32.6 ± 1.8 Average = 31.1 1 s = 0.6 95% CI = 4.9 IAEA-350 Tuna Distillation–INAA for organic Hg values reported as mg of Hg per gram dry mass 3.71 ± 0.03 3.79 ± 0.02 3.32 ± 0.02 3.63 ± 0.01 3.65 3.32–4.01 (95% CI) Average = 3.61 1 s = 0.20 95% CI = 0.69 Analyst, May 1998, Vol. 123 781Significant sources of uncertainty associated with this procedure include irradiation geometry differences between samples and standards (55% relative uncertainty, per sample), counting geometry differences between samples and standards (52% relative uncertainty) and measurement replication which includes the statistical uncertainty associated with counting statistics and the degree of material homogeneity. The relative contribution of this last component ranges from 1% to about 15% depending on the concentration of the analyte and the degree of material homogeneity.The total or expanded uncertainties for the analyses of these materials include the contributions from the above listed uncertainty components combined in quadrature and multiplied by a constant to yield values for the 95% confidence intervals,27 listed in Table 3. The results from the determination of methylmercury in SRM 2976 are given in Table 4. The uncertainties associated with counting statistics for each value are listed, together with the average, standard deviation of the average and the expanded uncertainty (U) which defines the 95% confidence interval.27 For this material, a small amount of selenium was present on most of the filter-papers, probably from the mussel tissue.Selenium accounted for 5 –14 % of the total area of the 279 keV gamma ray peak used to quantitate the amount of mercury from 203Hg. This interference was accounted for by using the countrate of the 264 keV gamma ray peak from the same isotope of selenium (75Se) and the measured ratio of the count-rates from the two selenium peaks, determined using an irradiated mercury-free selenium standard.Results for six of the seven aliquots are similar, with values ranging from 26.7 to 31.3 ng g21, but the seventh value of 19.9 ng g21 may be an outlier. This result does not appear to be an indication of any bottle-tobottle variation because the result of 26.7 ng g21 for the other aliquot from the same bottle is not significantly different from the other values.All the sample-to-sample variation in six of the values for SRM 2976 may be attributed to the uncertainty due to counting statistics alone. The cause of the one low value of 19.8 ng g21 is not known. However, this one apparently low value does affect the calculated average and standard deviation (1s) of the average. The calculated average and standard deviation for the seven aliquots is 27.9 ± 3.9 ng g21 and the corresponding value for six aliquots, 29.2 ± 1.8 ng g21.The results from analyses of SRM 2976 agree well with those reported in the literature (Table 5). It is interesting that the average value obtained from replicate analyses of one unlabeled bottle of SRM 2976 reported by the IAEA laboratory, 20.8 ng g21, is similar to the value obtained for one aliquot from bottle 950 (see Table 5). This could be an indication of material inhomogeneity.Additional work is necessary to determine the homogeneity of this material for a given portion size. Analytical results for methylmercury in SRM 2974 are also given in Table 4. Again, the uncertainties associated with counting statistics, the average, standard deviation of the average and the 95% confidence interval are listed. The values for five of the six aliquots are similar, ranging from 68 to 94 ng g21 but the seventh value of 180 ng g21 differs significantly and was not included in the average value or any other statistical evaluations of the data.It is possible that this sample or filter paper was contaminated. This is not an indication of a bottle-tobottle difference since the value obtained from another aliquot from the same bottle (68 ng g21) was not elevated. For SRM 2976, all sample-to-sample variations for six of the seven subsamples could be attributed to the uncertainty associated with counting statistics alone. This is not the case for SRM 2974; an additional 10% relative uncertainty is observed, beyond that associated with counting statistics.There was no measurable selenium interference present for SRM 2974 or IAEA-350. The results of analysis for SRM 2974 agree well with those reported in the literature (see Table 5). The standard deviation of the average value for five samples, 78.3 ± 9.6 ng g21 (standard deviation, 1s), is relatively large for this material but is similar to that obtained using the NIST GC–AED method8 (80.4 ± 8.4 ng g21). It is possible that the material is not homogeneous with respect to this analyte for subsample sizes up to 1 g.Subsample sizes ranged from 200 to 250 mg for this method and were approximately 1 g for the method described by Donais et al.10 Additional work is necessary to determine the homogeneity of this material for a given portion size. Conclusions Pre-irradiation acid distillation of organomercury compounds on to cysteine-loaded filter-papers followed by NAA for the determination of mercury was shown to be an adequate method for the determination of methylmercury in two mussel tissue and one fish tissue reference material.Results from replicate analyses of these materials showed that this method is reproducible and compares well with other methods. The average value and range of values determined using this method are similar to those reported by researchers using other methods. Table 4 Concentrations of organic mercury in two mussel tissue Standard Reference Materials expressed as ng Hg p of dry mass Material Measured value/ng g21 SRM 2976, bottle 2400 29.0 SRM 2976, bottle 1 31.2 SRM 2976, bottle 950 19.9 SRM 2976, bottle 1400 29.0 SRM 2976, bottle 350 31.3 SRM 2976, bottle 950 26.7 SRM 2976, bottle 2000 28.1 Range 19.9–3.13 Average ±1s 27.9 ± 3.9 Average ±95% CI 27.9 ± 5.2 SRM 2974, Bottle 219 67.6 SRM 2974, Bottle 246 80.2 SRM 2974, Bottle 362 93.5 SRM 2974, Bottle 362 74.3 SRM 2974, Bottle 320 75.8 SRM 2974, Bottle 219 (180)* Range 67.6–93.5 Average ±1s 78.3 ± 9.6 Average ±95% CI 78.3 ± 15.0 * Not included in the statistical analysis of the data.Table 5 Methylmercury concentrations (as ng Hg g21 dry mass) in SRM 2976 and SRM 2974 determined in this work and by three other laboratories.10,24 Uncertainties represent 1s. MeHg/ng g21 as Hg Laboratory (technique) SRM 2976 SRM 2974 NIST (GC–AED) 28.3 ± 2.2 80.4 ± 8.4 Research centre J�ulich (IEC–CVAAS) 28.1 ± 1.5 71.7 ± 1.5 IAEA (GC–CV–AFS) 26.8 ± 1.5 79.8 ± 3.2 IAEA (GC–CV–AFS) 20.8 ± 1.4* This work, NIST (distillation–NAA) 27.9 ± 3.9 78.3 ± 9.6 Certified value 27.8 ± 1.1† 77.3 ± 3.1‡ * Unpublished results from replicate analyses of subsamples from one sample; this value was not used for certification of SRM 2976.† The certified value is based on results given in this table with the exception noted above, and the uncertainty on this certified value represents a 95% confidence interval; see NIST Certificate of Analysis for Standard Reference Material 2976 Mussel Tissue, 1997.‡ The certified value for SRM 2974 is based on the results given in this table and results of analysis of the fresh frozen form of this same material, SRM 1974a; see NIST Certificate of Analysis for Standard Reference Material 2974 Organics in Freeze-Dried Mussel Tissue, 1997. 782 Analyst, May 1998, Vol. 123The results presented here for SRM 2974 Organics in Mussel Tissue (Freeze-Dried) and SRM 2976 Mussel Tissue were used in the process of assigning certified values for methylmercury concentrations for both materials.Preliminary experiments using this method for the determination of methylmercury in a soil material, containing inorganic mercury at levels that are 100 times greater than organic mercury levels, showed that for a soil matrix additional steps are needed to eliminate the inorganic mercury which distills over with any organomercury compounds. Additional research is in progress to determine the efficacy of this method for the determination of organomercury compounds in additional biological matrices. Certain commercial equipment, instruments and materials are identified in this paper in order to describe adequately the experimental procedures.Such identification does not imply endorsement by the National Institute of Standards and Technology, nor does it imply that the materials or equipment identified are necessarily the best available for the purpose.Contributions of the National Institute of Standards and Technology are not subject to copyright. References 1 Swensson, A., and Ulfvarson, U., Acta Pharmacol. Toxicol., 1968, 26, 273. 2 Suzuki, T., Miyama, T., and Katsunuma, H., Ind. Health, 1967, 5, 149. 3 The Biogeochemistry of Mercury in the Environment, ed. Nriagu, J. O., Elsevier/North-Holland Biomedical Press, Amsterdam, 1979, ch. 16–24. 4 Horvat, M., Bloom, and N.S., Liang, L., Anal. Chim. Acta, 1993, 281, 135. 5 Leermakers, M., Lansens, P., and Baeyens, W., Fresenius’ J. Anal. Chem., 1990, 336, 655. 6 Stoeppler, M., and Mathes, W., Anal. Chim. Acta, 1978, 98, 389. 7 Sellers, P., Kelly, C. A., Rudd, J. W. M., and MacHutchon, A. R., Nature (London), 1996, 380, 694. 8 Horvat, M., and Byrne, A. R., Analyst, 1992, 117, 665. 9 Emteborg, H., Bj�orklund, E., � Odman, F., Karlsson L., Mathiasson, L., Frech, W., and Baxter, D.C., Analyst , 1996, 121, 19. 10 Donais, M. K., Uden, P. C., Schantz, M. M., and Wise, S. A., Anal. Chem., 1996, 68, 3859. 11 J. Res. Natl. Inst. Stand. Technol., 1996, 101, 729. 12 Official Methods of Analysis of the Association of Official Analytical Chemists, AOAC, Philadelphia, 1995, Metals and Other Elements Ch. 9, AOAC Official Methods 983.20 and 988.11, Mercury(Methyl) in Fish and Shellfish, pp. 22–25. 13 Orvini, E., and Gallorini, M., in Special Publication 422, National Bureau of Standards, Gaithersburg, MD, and Proc. 7th IMR Symp., 1976, pp. 1233–1240. 14 Ebinghaus, R., Hintelmann, H., and Wilken, R. D., Fresenius’ J. Anal. Chem., 1994, 350, 21. 15 May K., Stoeppler M., and Reisinger K., Toxicol. Environ. Chem., 1987, 13, 153. 16 West�o�o, G., Acta Chem. Scand., 1967, 21, 1790. 17 Zelenko, V., and Kosta, L., Talanta, 1973, 20, 115. 18 Gvadjancic, I., Kosta, L., and Zelenko, V., Zh. Anal. Khim., 1978, 32, 812. 19 Horvat, M., May, K., Stoepler, M., and Byrne, A. R., Appl. Organomet. Chem., 1988, 2, 515. 20 Horvat, M., Byrne, A. R., and May, K., Talanta, 1990, 37, 207. 21 Horvat, M., Bloom, N. S., and Liang, L., Anal. Chim. Acta, 1993, 28, 153. 22 UNEP–WHO–IAEA, Reference Method for Marine Pollution Studies No. 46, United Nations Environment Programme (UNEP), Regional Seas Programme Activity Centre, Geneva, 1987. 23 Kratzer, K., Benes, P., and Spevackova, V., Int. J. Environ. Anal. Chem., 1994, 57, 91. 24 Donais, M. K., Saraswati, R., Mackey, E. A., Demiralp, R., Porter, B. J., Vangel, M., Levenson, M., Mandic, V., Azemard, S., Horvat, M., May, K., Emons, H., and Wise, S. A., Fresenius’ J. Anal. Chem., 1997, 358, 424. 25 Padberg, S., Burow, M., and Stoeppler, M., Fresenius’ J. Anal. Chem., 1993, 346, 686. 26 Curie, L. A., Anal. Chem., 1968, 40, 586. 27 Taylor, B. N., and Kuyatt, C. E., NIST Technical Note No. 1297, National Institute of Standards and Technology, Gaithersburg, MD, 1994. Paper 7/07453K Received October 16, 1997 Accepted December 16, 1997 Analyst, May 1998, Vo

ISSN:0003-2654    

DOI:10.1039/a707453k

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Kinetic approach to the chemical speciation of trace metals in soils† Alain Bermond*a, Isabelle Yousfib and Jean-Philippe Ghestema a Institut National Agronomique, Laboratoire de Chimie Analytique, 16 rue C. Bernard, 75231 Paris Cedex 05, France b CEA–IPSN/DPRE/SERGD/Laboratoire d’Etude du Stockage de Surface, BP 6, 92265 Fontenay aux Roses, France The reactivity of trace metals (Cd, Pb, Zn, Cu) present in two polluted soils was studied using two extractants, EDTA and hydroxylamine.As the classical chemical speciation of trace metals in soils when using these reagents seems to be unreliable (readsorption phenomena, non-selectivity of chemical reagents, soil/solution ratio, etc.), we applied two different kinetic approaches to the results in order to determine the speciation of trace metals in the two soil samples studied. The advantages of and problems with these approaches are discussed. Keywords: EDTA; hydroxylamine; trace metals; polluted soils; speciation; kinetics Regardless of their origin and the reasons for the increase in their concentration in soils, trace metals are liable to contaminate the food chain by migrating towards ground water or by accumulating in plants.This possible mobility and bioavailability are the result of the reactivity of trace metals in soils, in other words, their localization in different soil components, which is now usually called speciation. Note that, by also using the term ‘localization’, speciation of trace elements in soils is not fully defined by Florence,1 who first distinguishes the different physico-chemical forms of the same element.Here, in our discussion of trace elements, we shall use the terms ‘speciation’ and ‘localization’ interchangeably. Speciation of trace elements in soils may be performed using physical or chemical methods. In the first case, the possible methods2-4 are generally not sufficiently sensitive, and therefore can only be used for this purpose with severely contaminated samples.Chemical methods are more sensitive, and consist of using different chemical reagents (Table 1) for the extraction of trace elements, terminating with their quantification in the extraction phase (usually when equilibrium is reached).5 Several reagents are generally used in what are called sequential extraction protocols.6–11 However, it is now agreed that these protocols cannot supply a reliable estimate of the speciation of trace elements in soils,12–18 particularly for thermodynamic reasons (measurements made at equilibrium).This is why it appears necessary to consider other methods of determining this speciation of trace metals in soils in more detail, particularly considering kinetic aspects that also characterize the stability of the various trace metal–soil constituent associations. The first part of this paper includes some experimental results obtained with the use of two reagents for the extraction of trace metals in soils (hydroxylamine and EDTA), for what may be called the speciation equilibrium method.In the second part, we consider the possibilities of a kinetic approach, using the results obtained with the same two reagents, in order to determine the speciation of trace metals in soil samples. Experimental Soil samples Soil samples were taken from polluted sites. One of them, Couhins, is located in the Bordeaux region of France and forms part of a long-term agronomic test being carried out by the INRA Agronomy Unit on the consequences of spreading sewage sludge from a treatment station with a high concentration of Cd and Ni.The second, Evin, is located in the north of France and is part of an agrosystem contaminated by atmospheric industrial fallout around a metallurgical plant. Table 2 gives the main physico-chemical characteristics of the two samples (0–20 cm layer taken in 1994) measured on the granulometric fraction smaller than 2 mm after sieving.Total concentrations in this case were determined after mineralization with hydrofluoric acid. Extractions with hydroxylamine For the kinetic study of the extraction of metal cations by hydroxylamine, polyethylene tubes washed in acid and containing 1 g of fine soil (sieved to 2 mm and coarsely ground) and 50 ml of a solution of hydroxylamine hydrochloride (0.1 mol l21) were stirred using a rotary stirrer for a given mixing time, different for each tube and ranging from 30 min to 24 h.At the end of the chosen mixing time, the tube was removed from † Presented at The Third International Symposium on Speciation of Elements in Toxicology and in Environmental and Biological Sciences, Port Douglas, Australia, September 15–19, 1997. Table 1 Some chemical reagents used for the localization of soil trace metals Soil compartments Reagents attacked Chemical reactions Reducing reagents— NH2OH Oxalate Fe, Mn oxides Acidification and oxide reduction (dissolution) Oxidizing reagents— H2O2 NaClO4 Organic matter Oxidation and dissolution of organic matter Complexing reagents— Na4P2O7 EDTA Organic matter, Fe oxides, etc.Metals extraction by complexation Table 2 Some physico-chemical characteristics of the soil samples Sample source Parameter Couhins Evin pH 7.1 7.9 Organic matter (%) 2.2 1.8 Clay (%) 2.9 16.5 Cd (total)/µg g21 94.9 19.4 Zn (total)/µg g21 151 1415 Cu (total)/µg g21 45.3 43.5 Pb (total)/µg g21 44.8 1120 Analyst, May 1998, Vol. 123 (785–789) 785the stirrer, the pH was measured after decantation, the solution was filtered using Millipore (Bedford, MA, USA) filter membranes (pore diameter 0.45 mm, membrane type HVLP), and the filtrate was acidified to pH 2 by adding the necessary volume of concentrated nitric acid and kept at 4 °C until analysis. The isolated effect of the pH at equilibration time was also studied: when using dilute solutions of nitric acid for trace metal extraction, the effect of decreasing the redox potential alone can be deduced from comparison with trace metal concentrations extracted in the more or less acidified hydroxylamine medium (NH2OH + H+).A study of the amounts extracted at equilibrium as a function of pH was carried out using 1 g of soil and 50 ml of solution (nitric acid and/or hydroxylamine hydrochloride) at various pH values, with stirring as above for 24 h; the same protocol as described above was then used and the pH range studied was 2–6.To study the possible sorption of metal cations extracted during the reaction with hydroxylamine, we carried out an additional extraction after the extraction with this reducing reagent and using barium perchlorate, a reagent used by some workers to extract exchangeable trace metals.18,19 The sample was centrifuged between these two extractions; after separating the floating material, the bottom material was rinsed with water purified with a Milli-Q system (Millipore) and then allowed to react with a 0.5 mol l21 solution of Ba(ClO4)2 for 140 min.The remaining steps in preparation for the analysis were the same as above. The results presented here correspond to the results obtained for cations which were not extracted during a barium perchlorate extraction applied directly to the soil sample (without previous hydroxylamine reaction). Extractions with EDTA The study of the extraction of trace metals from soils by EDTA (Na2H2Y) was carried out at different concentrations of this reagent (0.0001–0.05 mol l21) over a wide range of pH values.Therefore, extraction solutions were acidified with nitric acid or were neutralized with sodium hydroxide. In order to determine the influence of acidity on the amounts of trace metals extracted, we also carried out extractions without EDTA at different pH values (calibration curves). For all extractions, 1 g of the soil sample was mixed with the extracting solution (50 ml) and stirred for 24 h.Solutions were filtered (filter pore diameter 0.45 mm) after the reaction to permit analysis of the extracted trace metals. A kinetic study of the extraction of metal cations by EDTA was made on a 19 g test sample and an initial volume of 950 ml of the extracting solution (0.05 m EDTA at pH 6.5), which corresponds to a soil/solution ratio of 1 : 50. Samples of 25 ml were processed for the determination of trace elements, as in the study at equilibrium.The various reagents used for this part of the study were of analytical-reagent grade. Determination of extracted cations All solutions were prepared from analytical-reagent grade salts or Titrisol solutions (Merck, Darmstadt, Germany) and Milli-Qpurified water. Cations in solution (Cd, Zn, Pb, Cu, Ni) and a few major cations (Fe, Mn, Mg, Ca) were measured with a Varian (Palo Alto, CA, USA) Spectra 250 Plus atomic absorption spectrometer using an air–acetylene flame and external standards.The analytical performances (sensitivity, detection thresholds) for the various extraction environments studied were not different from those obtained with simple matrices. Furthermore, we used the standard addition method to verify that there were no interferences and in all cases we made extraction blanks. Under these conditions, the relative standard deviations obtained for the extraction results with the reagents used, including the variability of the soil sample, were satisfactory and ranged between 2 and 6% (n = 4) depending on the metal.Estimate of kinetic parameters We adjusted the experimental curves (concentration of extracted cations as a function of time) using a non-linear regression program based on Marquardt’s algorithm and developed by the INA chemical laboratory.20 The calculation of the coefficients used in a given model was accompanied by the calculation of their confidence intervals and the various statistical parameters that evaluate the quality of the estimate.Results and discussion Studies at equilibrium Hydroxylamine Hydroxylamine is a reducing reagent that makes some of the Fe (and Mn) hydroxides in soils pass into solution, based on the following reaction: Fe2O3 (s) + 4H+ + 2NH2OH " 2Fe2+ + 5H2O + N2 (1) Fig. 1 shows that equilibria are reached for most metal cations after about 24 h. However, equilibrium seems slower for iron. The amounts of trace elements extracted after 24 h of reaction decrease when the pHf (the solution pH at the end of the experiment) increases.This is a classical result that involves the extracting power of H+ ions based on the following reaction: Si–M + 2H+ " SiH2 + M2+ (2) where Si represents an adsorption site of a constituent of the solid phase of the soil and M2+ is a metal ion that can be fixed to it. However, the effect of acidity is more complex since, as indicated by reaction (1), the protons also take part in the oxidation–reduction reaction and are therefore liable to modify the amounts of Fe dissolved.Monitoring the extracted cations as a function of time can give asymptotic curves for the two samples in this study, and therefore there is no reason to suppose that there are any refixation phenomena, as some workers have demonstrated using an almost identical reagent.19 The use of an additional extraction with barium perchlorate allowed us to establish that refixation phenomena occur during hydroxylamine reaction.21 As an example, it can be seen that the magnitude of the refixation phenomena depends on the final acidity (Fig. 2). This metal cation refixation phenomenon has already been demonstrated by a number of workers14,18,22 for other reagents, e.g., hydrogen peroxide, which are assumed to destroy a Fig. 1 Monitoring of extracted metal cations during the reaction with hydroxylamine (Couhins sample). 786 Analyst, May 1998, Vol. 123compartment of the soil. This phenomenon can be explained from a thermodynamic point of view by considering the soil as a multi-ligand system and by displacement of all equilibria towards the fixation of trace elements on the remaining compartments when one of them is destroyed.12 EDTA EDTA is a well known complexing reagent. It is capable of directly extracting metal cations from several soil compartments by a competitive classic complexation reaction, but also of dissolving some other compartments with which trace elements are associated, e.g., iron hydroxides or carbonates.In other words, EDTA is a non-specific reagent that cannot a priori lead to localization of trace elements in a particular soil compartment. It is nevertheless used in soil science in some sequential extraction protocols5 or as a reagent capable of forecasting bioavailability.23 Therefore, it is useful to make a complete study of the reactivity of trace metals in soils in the presence of this molecule. Results obtained when this reagent is used with two soil samples that we have already described have been presented and discussed elsewhere.24 As an example, the amounts of Pb (Evin sample) extracted by EDTA at different concentrations as a function of the pH of the solution are shown in Fig. 3. Finally, it can be seen that EDTA at a high concentration is a powerful means of extracting trace elements. Table 3 gives the amounts extracted: with all the studied cations included, this complexing reagent extracts between 40 and 90% of the total contents.Kinetic studies In this second part, we shall consider the advantage of using kinetic concepts to determine the speciation of trace elements in soils when this method, not frequently used nowadays in soil science,25 is applied to the results obtained with these two reagents. Hydroxylamine We have seen that a refixation phenomenon of a non-negligible part of trace metals took place during the reaction of a soil sample with hydroxylamine, although the shape of the curves of metal concentration as a function of time is always asymptotic. Schematically, trace metal extraction and refixation reactions may be represented by the following chemical reactions : k1 k2 A ––? B [| C (3) k3 where A = trace metal in iron hydroxide, B = trace metal in solution and C = trace metal refixed on other solid sites.The first reaction corresponds to cations going into solution following the dissolution of iron hydroxides and the second represents the refixation and dissolution equilibrium, if applicable, of trace metals on fixation sites in the sample (except for iron hydroxides).Note that from an experimental point of view, the only measurable concentration in our studies is the concentration of B measured in the solution. If it is assumed that all reactions are of first order, for simplification purposes, we can define a system of differential equations.For example, one of these equations is d[B]/dt = k1[A] 2 k2[B] + k3[C] (4) where the kinetic constants k1, k2 and k3 correspond to reaction (3). To use these data from a kinetic point of view, the solution of this type of system has been described in the literature,26 giving equations for [B] and [A] as a function of the three kinetic constants and for [A] and [C] at time t = 0, for which [B] = 0 by definition. We calculated the coefficients and concentrations involved using the non-linear estimation program mentioned in the Experimental section.The results of the statistical tests are generally satisfactory, and the example in Fig. 4 shows good agreement between the experimental and calculated curves for the concentration of B. Fig. 2 Amount of lead extracted with hydroxylamine or hydroxylamine and barium perchlorate as a function of pH (Evin sample). Fig. 3 Amount of lead extracted versus pH and EDTA concentration (Evin sample).Table 3 Amounts of EDTA-extracted trace elements at pH 6.5 Concentration/mg g21 (% of total) Metal Couhins sample Evin sample Cu 25 (55) 26 (60) Zn 77 (50) 560 (40) Cd 65 (69) 17 (74) Pb 33 (75) 1050 (93) Fig. 4 Calculation, according to eqn. (3), of amounts of lead (mg g21) bound to iron hydroxides [A] and extracted [B] and comparison with the experimental values [B] (Evin sample, final pH = 3.73). Analyst, May 1998, Vol. 123 787When the various coefficients are known, the variation of the concentration of A can be calculated for the same example as a function of time.This curve represents the concentration of cations related to the iron hydroxide compartment in the sample. According to this result, at time t = 0, the concentration A0 (the hydroxylamine-extractable amount of trace metals) is approximately 0.9 mg g21, in other words, about twice the value of the concentration of B when equilibrium is reached. The difference between these two values clearly corresponds to the refixed cations.Therefore, this kinetic approach is sufficient to determine the amounts of cations actually associated with the iron hydroxide compartment. In fact, considering the relatively high acidity of the reagent and the resulting non-selectivity (other compartments attacked by protons), it would be more accurate to state that this kinetic approach gives a better estimate of the amounts of metals in the soil sample concerned by attack with the acidified hydroxylamine reagent than can be obtained by measurement.EDTA The use of kinetic data for the extraction of trace metals from soils by EDTA cannot give the speciation (localization) of these elements in terms of soil compartment, as we have already seen. However, we studied the feasibility of a strictly operational speciation in which trace metals extracted by EDTA would be classified into labile metals (quickly extracted) and non-labile metals (less quickly extracted).This objective, together with simplifying assumptions about first-order reactions, requires that the constants of an equation of the following type should be estimated by non-linear regression: Q = C1exp(2k1t) + C2exp(2k2t) (5) where Q represents the amount of metals extracted at time t, C1 and C2 represent the labile and non-labile amounts, respectively, for a given metal and k1 and k2 are the kinetic constants associated with them. An example of the results obtained using this procedure and operational conditions is presented in Fig. 5. It can be seen that there is good agreement between the calculated and experimental curves; furthermore, the various statistical tests (correlation coefficient, confidence interval, residual variance) are also satisfactory. Identical results were found for other metals and other samples. Table 4 presents simple statistical parameters obtained during four repetitions (four independent samples) of this kinetic speciation method. In this case, C1 and C2 were calculated with respect to total metal concentrations. It can be seen that the proportion of the labile element is between 40 and 55% for the four metals studied.The values of the relative standard deviations ( < 10%) show that the repeatability of this method is satisfactory considering the variability of the samples. Therefore, it appears feasible to classify the extraction curves for trace metals in soils by EDTA into two metal categories, labile and non-labile.The benefits of doing so in terms of predicting the bioavailability and/or mobility of trace elements in the environment still have to be demonstrated. We shall simply present the initial results of this study that is now in progress. We have applied this method to freshly contaminated samples in the laboratory for which it is known that the added metals are very mobile, i.e., they should be classified in the labile category for the kinetic speciation method.Table 5 shows an example of the results obtained in which it is seen that added copper and cadmium are mostly found in the labile partition C1. Therefore, the various results appear to show that this operational method is useful and a more detailed study is under way. Conclusion Although chemical reagents are frequently used nowadays to extract and determine the localization or speciation of trace metals in soils, the two studies presented here on the extraction of trace metals at equilibrium show the limitations of this approach.The hydroxylamine example shows that trace metal refixation phenomena take place during the attack on the soil sample and consequently the measurements made at equilibrium underestimate the amount of metal cations related to the ‘reducible’ partition of the soil. Furthermore, a second problem occurs following the use of this reagent due to acidification of the hydroxylamine and the specific reactivity of protons concerning the extraction of trace metals.EDTA is necessarily a non-specific reagent since it can extract trace elements from several soil partitions. Therefore, measurements of trace elements extracted at equilibrium cannot be related to their speciation. A kinetic approach to the speciation of trace elements in soils (or sediments) may be an alternative way. The few examples shown in this work establish the advantage of this approach. Considering the results obtained with hydroxylamine, the use of kinetic data gives a more precise estimate of the amount of extracted trace metals than can be obtained by conventional measurements made when equilibria are reached, owing to refixation phenomena.Table 4 Statistical parameters used to characterize the repeatability of the ‘kinetic speciation’ of trace metals in soils Cd Zn Cu Pb Parameter C1 C2 C1 C2 C1 C2 C1 C2 Calculated values of C1 and C2/mg g21 55–56 51–49 19–23 21–23 41–41 39–37 16–19 16–16 41–41 38–38 23–21 21–21 45–41 41 28–29 31 Mean/mg g21 52 21.5 39.5 17 39.5 21.5 42 28.5 RSD (%) 5.2 9 5 9 4.3 4.6 5.4 2.5 Fig. 5 Monitoring of EDTA-extracted zinc as a function of time and comparison with the calculated curve (Evin sample; EDTA concentration = 0.05 m; pH = 6.5). 788 Analyst, May 1998, Vol. 123Kinetic data can be used differently when EDTA is used as a trace metal extraction reagent, to define the amounts of labile and non-labile cations for a given sample.If the results of this study (which still needs further work and validation) are satisfactory, it could contribute towards the creation of a tool for predicting risks to the environment due to the presence of trace metals in soil. References 1 Florence, J. M., Talanta, 1982, 29, 345. 2 Förstner, U., Int. J. Environ. Chem., 1992, 51, 5. 3 Charlet, L., and Manceau, A., Environ. Anal., Phys. Chem. Ser., 1993, 2, 118. 4 Manceau, A., Charlet, L., Boisset, M. C., Didier, B., and Spadini, L., Appl.Clay Sci., 1992, 7, 201. 5 Beckett, P. H. T., in Advances in Soil Science, ed. Steward, B. A., Springer, New York, 1988, pp.144-171. 6 Lake, D. L., Kirk, P. W. W., and Lester, J. W., J. Environ. Qual., 1984, 13, 175. 7 Quevauviller, P., Rauret, G., Muntau, H., Ure, A. M., Rubio, R., Lopez-Sanchez, J. L., Fiedler, H. D., and Griepink, B., Fresenius’ J. Anal. Chem., 1994, 349, 808. 8 Rauret, G., Rubio, R., and Lopez-Sanchez, J. F., Anal. Chim.Acta., 1994, 286, 423. 9 Rauret, G., Rubio, R., and Lopez-Sanchez, J. F., Int. J. Environ. Anal. Chem., 1993, 36, 69. 10 Tessier, A., Campbell, P. G. C., and Bisson. M., Anal. Chem., 1979, 51, 844. 11 Shuman, L. M., Soil Sci., 1985, 140, 11. 12 Bermond, A., Environ. Technol., 1992, 13, 1175. 13 Bermond, A., and Benzined, K., Water Air Soil Pollut., 1991, 57, 883. 14 Kheboian, C., and Bauer, C. F., Anal. Chem., 1987, 59, 1417. 15 Miller, W. P., Martens, D.C., and Zelazny, L. W., Soil Sci. Soc. Am. J., 1986, 50, 598. 16 Rendell, P. S., Batley, G. E., and Cameron, A. J., Environ. Sci. Technol., 1980, 14, 314. 17 Whalley, C., and Grant, A., in Heavy Metals in the Environment, ed. Allan, R. J., and Niagru, J. O., CEP Consultants, Toronto, 1993, pp. 286–299. 18 Nirel, P. M. V., and Morel, F. M. M., Water Res., 1990, 24, 1055. 19 Auliitia, T. U., and Pickering, W. F., Talanta, 1988, 35, 559. 20 Whittall, K. P., in Signal Treatment and Signal Analysis in NMR, ed. Rutledge, D. N., Elsevier, Amsterdam, 1996, pp. 46–49. 21 Yousfi, I., and Bermond, A., Environ. Technol., 1997, 18, 139. 22 Bermond, A., and Malenfant, C., Sci. Sol., 1990, 28, 43. 23 Juste, C., and Mench, M., in Biogeochemistry of Trace Metals, ed. Adriano, D., Lewis, Boca Raton, FL, 1994, pp. 159–165. 24 Ghestem, J. P., and Bermond, A., Environ. Technol., in the press. 25 Gutzman, D. W., and Langford, C. H., Environ. Sci. Technol., 1993, 27, 1388. 26 Romiguin, N. M., and Ropiguina, E. N., in Consecutive Chemical Reactions, ed. Scheider, R. F., Van Nostrand, Princeton, NJ, 1964, pp. 136–143. Paper 7/07776I Received October 28, 1997 Accepted January 5, 1998 Table 5 Labile (C1) and not labile (C2) cadmium and copper in freshly spiked samples Concentration/mg g21 (% of total) Fraction Cd Cu Labile fraction (C1) 35.8 (97.3) 50 (92.5) Non-labile fraction (C2) 1.0 (2.7) 4.0 (7.5) Analyst, May 1998, Vol. 123 789

ISSN:0003-2654    

DOI:10.1039/a707776i

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Bioavailability and speciation of arsenic in carrots grown in contaminated soil† Hans Helgesena and Erik H. Larsen*b a Department of Chemistry, Technical University of Denmark, Building 207, DK-2800 Lyngby, Denmark b Danish Veterinary and Food Administration, Institute of Food Research and Nutrition, 19 Mørkhøj Bygade, DK-2860 Søborg, Denmark, E-mail: ehl@vfd.dk Carrots were grown in seven experimental plots (A–G) containing mixtures of arsenic-contaminated and uncontaminated soil at concentrations ranging from 6.5 to 917 mg g21 (dry mass). The carrots harvested from plots A–D (6.5–338 mg g21 arsenic in the soil mixtures) showed a gradually increasing depression of growth with increasing level of contamination.At the experimental plots E–G with soil arsenic concentrations above 400 mg g21 no carrots developed. Whether this effect was caused by arsenic or the concomitant copper content which ranged from 11 to 810 mg g21 in the soil mixtures is unknown.The arsenic species extracted from the soils and carrots were separated and detected using anion-exchange HPLC coupled with ICP-MS. In the less contaminated soils from plots A and B arsenite (AsIII) was more abundant than arsenate (AsV) in the soil using 1 mmole l21 calcium nitrate as extractant. In the soils from plots C and D however, AsV dominated over AsIII whereas in the corresponding carrots AsV and AsIII were found at similar concentrations. Methylated arsenic species were sought after but not detected in any of the samples.The soil-to-carrot uptake rate (bioavailability) of arsenic was 0.47 ± 0.06% (average ± one standard deviation) of the arsenic content in the soils from plots A–D. In contrast to arsenic, the increasing copper content in the soils from plot A through D was not available to the carrots as the concentration of this element did not increase with increasing soil copper content. The ingestion of the potentially toxic inorganic arsenic via consumption of carrots grown in soil contaminated at 30 mg g21 in arsenic (plot B) was conservatively estimated at 37 mg week21.This was equivalent to only 4% of the provisional tolerable weekly intake (PTWI) for inorganic arsenic as suggested by the WHO and was therefore toxicologically safe. Consumption of carrots grown in more intensely arsenic-contaminated soils, however, would lead to a higher intake of inorganic arsenic and is therefore not recommended. Keywords: Arsenic speciation; carrots; plant uptake; soil contamination; high-performance liquid chromatography–inductively coupled plasma mass spectrometry Arsenic is a toxic element to humans and numerous studies have been conducted in order to assess the amount and chemical forms (species) of arsenic present in food and biological samples.The data generated have been used for evaluating whether ingestion of arsenic via consumption of food posed any health risk to humans.1 In nature, arsenic readily undergoes metabolic conversions mediated by microorganisms, plants and animals.2 This explains the finding in food of a range of arsenic species which possess different toxicity to humans.Modern sensitive and selective analytical techniques such as high-performance liquid chromatography (HPLC) coupled on-line with inductively coupled plasma mass spectrometry (ICP-MS) have made possible the separation and selective detection of arsenic species in food and environmental samples at their naturally occurring concentration levels.3 The concentration of arsenic in the terrestrial environment, including crop plants for human consumption, is generally low.4 In contrast, food items of marine origin contain arsenic at much higher concentrations, primarily as the non-toxic species arsenobetaine,5 whereas the toxic inorganic arsenic is present in seafood usually at a few per cent of the total arsenic content.6 Studies of the environmental contamination following industrial wood preservation have shown highly elevated concentrations of arsenic and copper in the environment near the source of contamination.These elements were detected in locally grown crop plants and in soil following atmospheric deposition.7 Furthermore, arsenic as arsenite (AsIII) and arsenate (AsV) were detected at several hundred micrograms per litre in the ground water8 sampled under the contaminated top soil. In order to evaluate the possible health risk to humans consuming crops cultivated in the contaminated soil, information is needed regarding the soil-to-plant uptake rate (bioavailability) of arsenic and speciation of the toxic element species in the crop plants.The bioavailability to crop plants of arsenic depends on several physical and chemical factors in the soil. The texture and chemical composition of the soil are important factors that govern the availability of arsenic to plants. Iron and aluminium oxides adsorb anionic arsenic species well in acidic soils, whereas calcium oxides in alkaline soils to a lesser extent adsorb anionic arsenic species.9,10 Anionic arsenic species are therefore in general more available to crops grown in alkaline than in acidic soil.11 Soil with a sandy texture is normally low in content of minerals and organic constituents which are capable of binding anionic arsenic species, and therefore a relatively high mobility of arsenic into the soil pore water may be observed.12 Phosphate in soil may compete with arsenate in its uptake by plants owing to the chemical similarities of the two anions.Thus, a low content of phosphate in soils may result in a high uptake of arsenate.13 Finally, the uptake of arsenic from soil to plant varies between plant species. In extreme cases this has led to an arsenic concentration in the edible mushroom species Laccaria amethystina sampled from arseniccontaminated soil14 at more than 1000 mg g21.In this case the arsenic was present in the mushroom as the low-toxic dimethylarsinic acid. In order to assess experimentally the bioavailability of arsenic from a contaminated soil to a crop plant, the selection of an extraction medium which simulates the plant-available fraction of the element is of importance. Extractants such as 0.1 † Presented at The Third International Symposium on Speciation of Elements in Toxicology and in Environmental and Biological Sciences, Port Douglas, Australia, September 15–19, 1997. Analyst, May 1998, Vol. 123 (791–796) 791mol l21 ammonium nitrate or ammonium acetate or calcium nitrate and water have been used for this purpose.15–18 However, in general the concentrations of the buffers used were higher than the natural ion strength of the interstitial water in soil. There is therefore a risk of overestimating the plantavailable amount of arsenic due to an unrealistically high desorption efficiency.The aim of this study was to investigate the uptake of total arsenic, arsenic species and copper by carrots cultivated in soils at different levels of contamination by these elements. Furthermore, the possible implications to human health associated with consumption of the contaminated carrots were considered. Carrot (Daucus carota) was selected as the test crop owing to its high tolerance to water soluble arsenic19 and because it is a commonly consumed vegetable. Experimental Design and conduction of field experiments The soil at a former industrial wood preservation site, located near Hillerød, Denmark, was contaminated by arsenic and copper because of spills that had occurred during impregnation of the wood by a mixture of arsenic pentaoxide, copper(ii) oxide and chromium trioxide.7 The arsenic content in the soil varied from 500 to 2000 mg g21 at particularly contaminated ‘hot spots’ according to a characterisation carried out by Frederiksborg County Environmental Administration.In order to establish a field experiment, soil was sampled from such a contaminated ‘hot spot’ and from an uncontaminated site nearby. The two soils were mixed using a cement mixer at a variety of ratios and included an uncontaminated soil (experimental plot A) and soils contaminated at gradually increasing levels of arsenic and copper (experimental plots B–G). The contaminated soil was a loamy sand type which was low in content of organic matter,18 whereas the uncontaminated soil was a loamy sand type rich in organic matter.Consequently, the mixtures of the two soils were loamy sand types rich in organic matter and loose in structure at plots A–D whereas the soils at plots E–G were denser owing to an increasingly higher fraction of the contaminated soil. The soil mixtures were filled into poly(vinyl chloride) (PVC) cylinders of 0.75 3 0.6 m id, which were placed vertically in holes dug in the ground at the uncontaminated site where the field experiment was conducted.The rim of the cylinders reached 10 cm above the surrounding uncontaminated soil. A perforated PVC pipe of 1 3 0.1 m id was placed in the soil at each experimental plot to facilitate drainage of water. Seeds of carrot (Daucus carota) were sown in the experimental soils. During growth, the experimental plots were sufficiently irrigated but neither fertilisers nor pesticides were used. After 17 weeks of growth the crops from experimental plots A–D were harvested whereas no carrots were obtained from plots E–G owing to strongly impaired growth or failure of the crops.The green tops of the carrots including 1–2 cm of the root were removed. The roots were rinsed thoroughly to remove all visible soil particles in order to prevent soil contamination. The carrots were separated into peel and core, and the samples were shredded and freeze-dried. The dried carrot samples were homogenised in a mortar to pass a 0.5 mm mesh sieve, and were stored dry in a desiccator until the time of chemical analysis.A representative soil sample (2 l volume) was taken from each experimental plot by mixing sub-samples taken by a soil drill. The soil samples were oven-dried at 60 °C for 3 d followed by passage through a 0.5 mm mesh sieve. Chemical analysis of arsenic and copper in carrots and soil A sub-sample of carrot (0.5 g dry mass) was wet ashed using 4 ml of sub-boiling distilled nitric acid in a DAE II Teflon-lined pressure steel bomb (Berghof, Tübingen, Germany) at 160 °C for 4 h.After cooling, the residue was diluted to 20 ml with water produced in a Super-Q apparatus (Millipore, Milford, MA, USA) prior to arsenic and copper determination as described in more detail elsewhere.20 A sub-sample of soil (1.0 g dry mass) was mineralised following a standard method of analysis21 in a 50 ml pressureresistant flask, and 20 ml of a mixture of nitric acid and water (1 + 1 v/v) was added.The Teflon-lined lid of the flask was tightened and the flask was placed in an autoclave for 1 h at 125 °C. After cooling, the acid–soil mixture was diluted to 100 ml with water. The total arsenic and copper contents of the diluted acidic residues of the carrot and soil samples were determined by Zeeman-effect electrothermal atomic absorption spectrometry, using atomisation from a graphite platform in pyrolytic graphite-coated graphite tubes (Zeeman 3030, Perkin-Elmer, Norwalk, CT, USA).The temperature and time programming and other settings used are given in Table 1. Quantification was based on peak area measurements using the method of standard additions for calibration as described in detail elsewhere.20,22 The total arsenic and copper contents determined in the soils and in the carrots are given in Tables 2 and 3. Extraction of arsenic species from soils and carrots The dried soil sample (5.0 g) was transferred into a 50 ml Erlenmeyer flask and 25 ml of a 1 mmol l21 calcium nitrate solution were added.The extraction of arsenic proceeded for 1 h with gentle mechanical shaking at room temperature. The supernatant was separated from the soil solids by centrifugation prior to injection into the HPLC–ICP-MS system for arsenic speciation determination. The arsenic contained in the dried carrot sample (100 mg) was extracted for speciation determination in 5.00 ml of methanol–water (1 + 9 v/v) using a microwave-assisted technique.The mixture, which was filled into a capped 10 ml centrifuge tube, was positioned in the focused microwave field of the microwave apparatus (Maxidigest MX 350, Prolabo, Paris, France). Prior to applying the microwave energy, 100 ml of cold water (ballast water) which surrounded the capped centrifuge tube holding the sample were added. Part of the microwave energy was absorbed by the ballast water, thus protecting the sample from overheating.In order to optimise the extraction efficiency, the microwave apparatus was operated at a range of power and time combinations. The optimum extraction efficiency was achieved using four treatments at 75 W power (25% of total power) for 8 min each, as shown in Fig. Table 1 Instrumental settings for Zeeman-effect atomic absorption spectrometric analyses Element Parameter Dry Dry Ash Atomise Clean Graphite furnace— Arsenic Temperature/°C 100 130 1150 2300 2600 Ramp/hold time/s 5/20 5/50 20/30 0/5 1/3 Copper Temperature/°C 100 130 1200 2200 2600 Ramp/hold time/s 5/20 10/40 20/30 0/5 1/5 Ar flow rate/ml min21 300 300 300 0 300 Spectrometer— Sample volume 20 ml Chemical modifier 10 ml of a mixture containing 1000 mg l21 palladium and 2000 mg l21 magnesium nitrate was used for the arsenic determinations Resonance wavelength 193.6 nm (arsenic) and 324.8 nm, (copper) Spectral bandpass 0.7 mm 792 Analyst, May 1998, Vol. 1231. After each of the treatments the sample tube was cooled under running tap water and the ballast water was renewed. The clear supernatant was injected into the HPLC–ICP-MS system without any further sample pre-treatment.Arsenic speciation by HPLC–ICP-MS The use of HPLC for the separation and ICP-MS for the selective detection of arsenic for speciation studies in biological samples has been extensively described elsewhere.8,20,23 In summary, the anionic arsenic species in the sample extracts were separated using an ION 120 organic polymeric strong anion-exchange HPLC column (Interaction Chromatography, San Jose, CA, USA), which was eluted isocratically at 1 ml min21 with 45 mmol l21 ammonium carbonate solution in water–methanol (97 + 3) at pH 10.3 as the mobile phase.The eluate from the HPLC system was continuously introduced into the ICP-MS instrument (Elan 5000, Perkin-Elmer SCIEX, Thornhill, ON, Canada), which was adjusted to monitor the 75As signal intensity at m/z 75 versus time.Aqueous mixtures of AsIII, AsV, monomethylarsonate (MMA) and dimethylarsinate (DMA) were injected into the anion-exchange HPLC–ICP-MS system and their retention times (tr) were recorded. In this way the chromatographic peaks emerging after the injection of the sample extracts were identified by their retention times as indicated in Fig. 2. The intensity of each chromatographic peak was quantified against corresponding calibration curves constructed by injection of standard mixtures of known concentrations into the HPLC–ICP-MS system.Results and discussion Arsenic and copper in soils and carrots A pronounced depression of the growth of the carrots was observed with increasing contamination of the soils of the experimental plots, shown by a decrease in the height of the carrot tops at experimental plots A–D. Furthermore, at plots C and D the carrot tops were wilted and partly yellow. The depression of growth was further illustrated by the average lengths of the harvested carrots, which were approximately 12, 10, 8 and 5 cm at plots A, B, C and D, respectively. No crop was obtained from experimental plots E–G, at which the soil arsenic concentrations (average ± one standard deviation of duplicate determinations) were 406 ± 30, 679 ± 95 and 917 ± 59 mg g21, respectively.The observed failure of the crops may have been caused by a phytotoxic effect of arsenic19 or by copper. The arsenic and copper contents in the soil at experimental plot A (Table 2) were similar to those found in uncontaminated Danish soils.24 Therefore, this plot was well suited as a reference for the results obtained from the contaminated experimental plots B–D.The results in Tables 2 and 3 show that the concentration of arsenic in carrots increased with increasing arsenic concentrations in the soils. Furthermore, the data in Table 3 show that the arsenic concentration in the peels was Table 2 Arsenic, arsenic species and copper in experimental soils* Extractable arsenic Experimental Extraction plot Total As† AsIII AsV efficiency (%) Total Cu† pH‡ A 6.5 ± 0.3 0.010 0.002 0.20 11.0 ± 0.6 4.4 B 30.0 ± 1.5 0.018 0.012 0.10 39.9 ± 2.0 5.2 C 93.3 ± 4.7 0.007 0.062 0.07 125 ± 6.3 4.7 D 338 ± 17 0.040 0.241 0.08 251 ± 13 5.0 Danish monitoring Median 3.6 7.0 values§ 95th percentile 8.4 15.9 * All concentration values are given in mg As g21 (dry mass).† Values given as mean ± one standard deviation of duplicate determinations.‡ Determined electrometrically after extraction using 0.01 mol l21 of calcium chloride in water. § Data from Jensen et al.24. Table 3 Arsenic, arsenic species and copper in carrots grown in experimental soils* Extractable arsenic Experimental Extraction plot Sub-sample Total As AsIII AsV efficiency (%) Total Cu A Core < 0.098 < 0.020 < 0.077 n.d 4.27 ± 0.21 Peel < 0.098 < 0.020 < 0.077 n.d 7.64 ± 0.38 B Core 0.112 ± 0.006 0.025 ± 0.007 < 0.077 n.d 2.93 ± 0.15 Peel 0.246 ± 0.012 0.038 ± 0.002 < 0.077 n.d 7.60 ± 0.38 C Core 0.387 ± 0.019 0.073 ±0.003 0.105 ± 0.012 46 ± 4 3.36 ± 0.17 Peel 1.04 ± 0.05 0.210 ± 0.048 0.261 ± 0.050 46 ± 10 3.20 ± 0.16 D Whole 1.85 ± 0.09 0.610 ± 0.033 0.672 ± 0.048 69 ± 4 5.21 ± 0.26 Danish monitoring Median < 0.045‡ 4.20§ values† 90th percentile < 0.045‡ 5.98§ * All concentration values are given as mean ± one standard deviation of duplicate determinations in mg As g21 (dry mass).† Corresponds to whole carrots. ‡ Data from National Food Agency of Denmark. § Data from Hansen and Andersen.25 Analyst, May 1998, Vol. 123 793higher than that in the core of the carrots by a factor of approximately three. In addition to arsenic, the soils were also contaminated with copper, as shown by the results in Table 2. The contents of copper in the corresponding carrots in Table 3 show that the concentration of this element did not change with a 23-fold increase in soil copper concentration from plot A to plot D.The copper content in the carrots from all experimental plots was within the normal range for uncontaminated carrots available on the Danish market,25 as indicated in Table 3. Hence, the high concentration level of copper in the contaminated soils was not bioavailable to the carrots, and further chemical investigations were therefore not considered. Selection and optimisation of extraction methods for arsenic species in soils and carrots Procedures reported in the literature for extraction of arsenic from soil15,16,18 often make use of physico-chemical conditions (pH, ionic strength) that deviate from those existing in natural soils.Thereby, there is a risk of over-estimating the fraction of arsenic which is soluble and available to the plant. Therefore, the requirements laid down in this study were that the extractant should approximate the ion strength and pH value of the interstitial water in the soils studied.To meet these requirements, an aqueous solution of calcium nitrate was used18 at a concentration of 1 mmol l21, which approximated the natural conditions in typical Danish soils.24 Furthermore, this extractant did not possess any redox or acid–base properties per se, which was of importance for the conservation of the original abundance of the AsIII–AsV redox pair during the extraction process. Furthermore, the extraction had to be carried out using a minimum of physical agitation to prevent the introduction of unrealistically high extraction forces to the system.Otherwise there was a risk of over-estimating the plant-available fraction of arsenic in the soils. The results in Table 2 show that the extractable arsenic in the soil was 0.07–0.2% of the total arsenic as estimated by leaching with hot nitric acid. The selection and optimisation of the extraction procedure of arsenic species from the carrots were aimed at obtaining a high extraction efficiency and at the same time conserving the arsenic species contained in the plant material.The application of continuous low-power microwave energy is advantageous for this purpose because the technique makes short extraction times possible with a reproducible and adjustable input of energy.26 Water was chosen as extractant and methanol (10% v/ v) was added to eliminate the risk of microbial growth. The microwave-assisted extraction gave an extraction efficiency of arsenic of 69%, as shown in Fig. 1. With these experimental conditions the temperature of the extracts did not exceed 70 °C. The conservation of the arsenic analytes was confirmed by the recovery of arsenic as AsIII spiked at 22 or 110 ng and as AsV spiked at 39 or 197 ng to separate sub-samples of the dry carrot material from plot A. The chromatographed extracts showed that no conversions of the two spiked arsenic species had occurred. Arsenic speciation in soils and carrots The speciation results in Table 2 show that AsIII and AsV were present in all soils.In the soils from plots C and D, AsV was present at a higher concentration than AsIII, whereas at plots A and B AsIII dominated. Historically, the arsenicals used for the industrial wood preservation process were predominantly the pentavalent sodium arsenate(v) and arsenic(v) oxide.18 The results obtained therefore indicate that a partial reduction to the trivalent form had occurred, possibly facilitated by the physicochemical or microbial conditions in the soil.The soil samples from plots A and B additionally contained a trace of the cationic arsenical trimethylarsine oxide (TMAO), which was detected in the extracts by cation-exchange HPLC–ICP-MS.20 Microbial activity may explain the finding of TMAO in these two soil samples.2 The uptake of the hydrogenarsenate ion by the carrot is assumed to follow the same route as the chemically similar hydrogenphosphate ion,12 whereas the uptake by carrots of AsIII, which is not ionised at the pH value of the soil, is unlikely.The results in Table 3, however, show that there are about equal concentrations of arsenic as AsIII and AsV in the carrot samples. These results suggest that following uptake by the carrot, AsV has been partially reduced to AsIII by bacteria associated with the root hairs or by the carrot itself. No methylated arsenic species were found in the carrots.Bioavailability of arsenic from soils to carrots The results in Table 4 show that the availability of arsenic to carrots (whole) from plots A–D is 0.47 ± 0.06% (RSD 13%) of Fig. 1 Extraction efficiency of arsenic from carrot against number of repeated microwave-assisted extraction treatments applied. See Experimental for details. Fig. 2 Anion-exchange HPLC–ICP-MS of A, an extract of carrot from experimental plot D and B, a mixture of four anionic arsenic species at DMA 8.6, AsIII 11.9, MMA 8.7 and AsV 21.3 ng As ml21.Volume injected, 100 ml. See Experimental section for details. 794 Analyst, May 1998, Vol. 123the arsenic concentration in the corresponding soils. The bioavailability of arsenic to the carrots expressed relative to the extractable arsenic in the corresponding soils is 580 ± 150% (RSD 25%). The higher RSD value of the latter estimate of the bioavailability is partly due to the inclusion of the extraction efficiency in these calculations, which contributes to the overall uncertainty.The former method of estimation of bioavailability is more practical because it only involves the relatively straightforward determination of total arsenic. For the discussion of the impact of ingested arsenic on human health, however, the speciation data in Table 3 are of great importance. Human health risk considerations Inorganic arsenic has been recognised as a human carcinogen which may cause skin or lung cancer.27 The fact that arsenic in the carrots is present as the toxic arsenic species calls for an evaluation of the safe use of arsenic-contaminated carrots for human consumption.A provisional tolerable weekly intake (PTWI) for inorganic arsenic has been established as 15 mg kg21 body mass from all food sources including water. There is only a small margin between the PTWI value and the adverse effects observed in epidemiological studies.28 This intake value should therefore not be exceeded. The arsenic concentration in the soil at experimental plot B (30 mg g21) is close in value to the soil quality criterion for total arsenic set by the Danish Environmental Protection Agency (EPA)24 at 20 mg g21.Soils contaminated by arsenic above this concentration level are not recommended for sensitive use such as cultivation of vegetables for human consumption. The results in Table 3 show that the carrots harvested from the same experimental plot B contain arsenic at 0.12 mg g21 (weighed average of core and peel) or 0.014 mg g21 on a fresh mass basis.Furthermore, the speciation results in Table 2 show that the arsenic in the carrots is present as inorganic species. A recently conducted food intake study29 showed that the adult Danes consume (90th percentile) up to 376 g (fresh mass) of vegetables per day. Assuming conservatively that this vegetable consumption rate is represented solely by carrots with an arsenic content equal to that found in carrots harvested from experimental plot B, the intake of inorganic arsenic amounts to 37 mg week21 or 4% of the PTWI value.A fraction of the arsenic content, however, remained unaccounted for with respect to its speciation because of the incomplete extraction efficiency as indicated in Table 3. Assuming that all arsenic in the carrot is present as inorganic species, the estimated intake is sufficiently low to allow for a contribution by inorganic arsenic from other food sources and water1,4 without any risk of exceeding the PTWI value for inorganic arsenic.It is therefore concluded that the 20 mg g21 soil quality criterion for arsenic established by the Danish EPA is sufficiently safe to prevent any unacceptable intake of inorganic arsenic via consumption of carrots. We thank Senior Advisor Poul Aaboe Rasmussen of the Frederiksborg County Environmental Administration for supporting the planning and implementation of the field experiments, and Professor Elo H.Hansen of the Technical University of Denmark for academic supervision (H.H.) and for the use of laboratory facilities. References 1 Larsen, E. H., PhD Thesis, National Food Agency of Denmark, Søborg, 1993. 2 Cullen, W. R., and Reimer, K. J., Chem. Rev., 1988, 89, 713. 3 Larsen, E. H., Fresenius’ J. Anal. Chem., 1995, 352, 582. 4 Food Monitoring in Denmark, Nutrients and Contaminants 1983–1987, Publication No. 195, National Food Agency of Denmark, Søborg, 1990. 5 Francesconi, K. A., and Edmonds, J. S., Oceanogr. Mar. Biol. Annu. Rev., 1993, 31, 111. 6 Edmonds, J. S., and Francesconi, K. A., Mar. Pollut. Bull., 1993, 26, 665. 7 Larsen, E. H., Moseholm, L., and Møller, M. M., Sci. Total Environ., 1992, 126, 263. 8 Larsen, E. H., Spectrochim. Acta, Part B., in the press. 9 Woolson, E. A., Axley, J. H., and Kearney, P. C., Soil Sci. Soc. Am. Proc., 1971, 35, 101. 10 Wauchope, R. D., and McDowell, L. L., J. Environ.Qual., 1984, 13, 499. 11 Frost, R. R., and Griffin, R. A., Soil. Sci. Soc. Am. J., 1977, 41, 53. 12 Atkins, M. B., and Lewis, R. J., Soil Sci. Soc. Am. Proc., 1976, 40, 655. 13 Woolson, E. A., Axley, J. H., and Kearney, P. C., Soil Sci. Soc. Am. Proc., 1973, 37, 254. 14 Larsen, E. H., Hansen, M., and Gössler, W., Appl. Organomet. Chem., in the press. 15 Harper, M., and Haswell, S. J., Environ. Technol. Lett., 1988, 9, 1271. 16 Räisänen, M. L., Hämäläinen, L., and Westerberg, L.M., Analyst, 1992, 117, 623. 17 Hlavay, J., Polyak, K., Bodog, K., and Csok, Z., Microchem. J., 1995, 51, 53. 18 Ottosen, L. M., PhD Thesis, Technical University of Denmark, Lyngby, 1995. 19 Grant, C., and Dobbs, J., Environ. Pollut., 1977, 14, 213. 20 Larsen, E. H., Pritzl, G., and Hansen, S. H., J. Anal. At. Spectrom., 1993, 8, 1075. 21 Danish Standard 259, DS Handbook 21.1., 1st edn., Dansk Standardiseringsråd, Copenhagen, 1991. 22 Larsen, E. H., J. Anal. At. Spectrom., 1991, 6, 375. 23 Larsen, E. H., and Stürup, S., J. Anal. At. Spectrom., 1994, 9, 1099. 24 Jensen, J., Bak, J., and Larsen, M. M., Heavy Metals in Danish Soils, Report 1996/4, National Environmental Research Institute, Roskilde, 1996. 25 Hansen, H. H., and Andersen, A., Lead, Cadmium, Copper and Zinc in Fruit and Vegetables 1977–80, Publication No. 84, National Food Agency of Denmark, Søborg, 1983. Table 4 Bioavailability of arsenic in carrots Ratio of concentrations of arsenic Ratio of concentrations of arsenic in carrot in carrot to arsenic in soil (%) to arsenic in soil extracts* (%) Experimental plot Core Peel Whole Core Peel Whole A n.d. n.d. 0.43† n.d. n.d. n.d B 0.37 0.82 0.41‡ 370 820 410‡ C 0.41 1.10 0.47‡ 590 1600 680‡ D 0.55 640 Mean ± SD 0.39 ± 0.03 0.93 ± 0.21 0.47 ± 0.06 480 ± 160 1200 ± 550 580 ± 150 * Data for extraction efficiency from Table 2. † Estimated from Danish monitoring data.4,25 ‡ Weighed average based on a 10 : 1 mass ratio of core and peel, respectively. Analyst, May 1998, Vol. 123 79526 Szpunar, J., Schmitt, V. O., Donard, O. F. X., and Lobinski, R., Trends Anal. Chem., 1996, 15, 181. 27 International Agency for Research on Cancer, IARC Monographs on the Evaluation of Carcinogenic Risk of Chemicals to Humans—Some Metals and Metallic Compounds, Vol. 23, IARC, Lyon, 1980. 28 World Health Organization, Toxicological Evaluation of Certain Food Additives and Contaminants, Food Additives Series, No. 24, WHO, Geneva, 1989. 29 National Food Agency of Denmark, Danskernes Kostvaner 1995, Hovedresultater, Publication No. 235, National Food Agency of Denmark, Søborg, 1996 (in Danish with summary in English). Paper 7/08056E Received November 10, 1977 Accepted February 27, 1998 796 Analyst, May 1998, Vol. 123

ISSN:0003-2654    

DOI:10.1039/a708056e

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Application of flow injection analysis adsorption–elution protocols for aluminium fractionation† Kipton J. Powell Department of Chemistry, University of Canterbury, P. Bag 4800, Christchurch, New Zealand FIA (flow injection analysis) is a widely used technique for trace element analysis, but has a number of inherent problems. When used to determine the ‘free’ metal concentration, the ‘normal’ reaction time in the manifold, 15–30 s, promotes a significant sequestration of metals from labile and pseudo-labile complexes. Also there is potential for matrix components other than the target analyte to affect the rate of the analyte–reagent reaction. This may lead to an over- or under-estimate of the amount of labile metal relative to calibrations based on simple aqueous standards. These problems can be minimised by use of flow systems with much shorter ‘reaction times’ and by separating the analyte fractionation step from the analyte–reagent reaction step.This can be achieved by use of on-line ‘adsorbents’ from which the captured analyte is eluted prior to the analyte–reagent reaction. This is illustrated for the fractionation of Al by use of ca. 1 s contact time with an oxine-derivatised gel. Real-time analysis of non-retained fractions coupled with selective elution of retained species provides the concentrations of three fractions: ‘free Al’ [Al3+ + Al(OH)2+ + Al(OH)2 +], ‘labile organic Al’ and ‘Al13’ hydroxy polymers. Quantitative separation of Al and Fe is achieved. For ‘free Al’ the linear working range is 0.3–16 mM, the LOD 70 nM and the RSD at 2 mM Al is 3.7%.The method is compared to conventional FIA for determination of ‘reactive Al’ in soil solutions and is applied to the Al-complexation capacity and pHdependent Al binding of a fulvic acid, and in the correlation of plant growth with Al fractions in soil solutions. Keywords: Flow injection; aluminium; speciation; 8-hydroxyquinoline; chrome azurol S; lucerne; fulvic acid; complexation capacity; soil solution The aquatic environment is a multicomponent, multiphase system. The components combine to give a diverse range of species which may differ in physical form, stoichiometry or oxidation state. The component concentrations may be fixed or variable. Species may be in rapid equilibrium (e.g., Cu2+–Cl2) or slow equilibrium (e.g., Cu2+-fulvic acid)1 or they may be inert (e.g., Bi–EDTA).Metals exist in a number of ‘pools’, e.g., free metal ions, hydrolysed metal ions, low molecular mass complexes (inorganic and organic), humic complexes and colloids. Within each pool a range of species may co-exist for a single metal, their relative concentrations being a function of pH and component concentrations. Their absolute concentrations are a function of the concentrations of all ligands and other metal ions. This paper focuses on speciation of metals in the aqueous phase, with particular reference to Al. The toxicity of a metal and its mobility are closely linked to its speciation. It is a generalisation that for metals which exist in nature in only one oxidation state the ‘free metal ion’ is often the most toxic.When a metal is bound to ligands in stable hydrophilic complexes the toxicity is suppressed because of diminished availability to micro-organisms.1 Both thermodynamic (stability) and kinetic (dissociation rate) factors determine the availability of a metal at the cell wall. In contrast, hydrophobic complexes can be very toxic,1 even if kinetically inert, because of direct penetration of the cell lipid bilayer. To the analytical chemist the challenge is to achieve a fractionation of the metal which targets only one species in the system. This can be achieved by use of a non-invasive probe (e.g., an ion-selective electrode) but these are available for few metal ions and the working ranges barely encompass that which is critical in environmental systems.Another option2 is based on the determination of the total amount of all elements in a sample followed by the computer-aided calculation of the equilibrium concentrations of species, based on metal complex and redox equilibrium constants, kinetic factors, adsorption and heterogeneous processes. This task is daunting and although comprehensive tabulations of stability constants are now available,3 adjusting constants to the correct ionic strength and temperature requires additional data or approximations. The alternative is kinetic-based analyses which involve a selective reaction with one species before significant reequilibration can occur. This is not difficult if the species formed by a given component are non-labile (e.g., SeIV and SeVI oxyanions). But for labile or moderately labile species it requires a very short experimental time scale.Anodic stripping voltammetry (ASV) has been used successfully to differentiate species which are ‘labile’ or ‘non-labile’ on the ASV timescale of, say, 100–500 ms. However this technique is limited to those metals which are reducible at the Hg electrode, which amalgamate with Hg and which strip reversibly. This is a comparatively small, though important, group of elements. Flow injection analysis: advantages The FIA (flow injection analysis) procedure allows highly reproducible and comparatively short (15–30 s) experimental times, is applicable to a wide range of metal analyses and can be coupled to a range of detection systems (e.g., spectrophotometric, amperometric). For Al it also provides a direct method of analysis, in contrast to methods such as Driscoll’s4 in which the reactive Al species (retained on a cation-exchange column) are determined as the difference between two other measurements, one of which involves an operationally defined fraction.Flow injection analysis: limitations for fractionation procedures One disadvantage of FIA is that the reaction time (throughout which fractionation and re-equilibration in the sample are occurring) is too long, except for slowly labile systems. The rate constants for the exchange of water from simple aqua ions indicate that only Al3+ has a half life of the order of seconds, although metal complexes with polydentate ligands are known to dissociate more slowly than the aqua ions.5,6 † Presented at The Third International Symposium on Speciation of Elements in Toxicology and in Environmental and Biological Sciences, Port Douglas, Australia, September 15–19, 1997.Analyst, May 1998, Vol. 123 (797–802) 797The long reaction coil used in a FIA manifold (Fig. 1) is required to achieve sample–reagent homogeneity in the sample zone. Improved mixing in a shorter time can be achieved by replacing the coil with a series of mixing tanks. Hawke and Powell7 found that 5–7 s was a practical lower limit with three mixing tanks, but this elapsed time was still sufficient for significant sequestering of Al (15–35%) from its malonate and oxalate complexes. The metallochromic reagents typically used for FIA determination of reactive Al, pyrocatechol violet (PCV), chrome azurol S (CAS) and eriochrome cyanine R (ECR), are very strong complexing agents.8–10 Given adequate time they will almost quantitatively remove Al from its complexes with simple organic ligands10 and substantially remove it from its humate complexes.7 The reaction rates of these reagents with Al3+ are strongly dependent on ionic strength at I < 0.1 m,7 a disadvantage when matrix matching of natural waters and soil solutions is not possible.Further, the pH of the buffer used with the reagent (Fig. 1) may be very different from that of the unknown (e.g., PCV uses a buffer at pH 6.2–6.5, yet samples from acidified environments will have pH < 5.4). This will cause a scrambling of the metal speciation during the reaction time. CAS uses a pH 5.0 buffer which is within the ‘acidified sample’ pH range and is optimal in this respect.In conventional FIA the fractionation process (in which the reagent targets one or more species) and the analytical process (formation of an analyte–reagent product) are combined. Any potential interferents should be masked, e.g., Fe3+ in the case of Al3+ analyses. If this involves the addition of a complexing agent then this may also affect the rate of release of Al from its complexes, or the rate of the Al–reagent reaction.11 The addition of ascorbic acid– bipyridyl to reduce FeIII to FeII has the effect of dissolving Fe–hydroxy colloids and releasing codeposited Al.7 It is important that the apparent rate constant for the analyte– reagent process be the same in the standards and the samples.If the kinetics of this reaction are sensitive to components in the sample medium then this will reflect as an error in the apparent fractionation. There is now ample evidence that this is the case for Al–metallochromic reactions. Simpson et al.11 observed that the rate of reaction of PCV with Al in the presence of different ligands was: oxalate Å F2 Å malonate > salicylate >> no ligand >> citrate. They also observed that (i) the rate of reaction of Al in humic samples decreased with sample dilution (i.e., with decrease in [organic ligands]), and (ii) in the presence of humic substances the rate of reaction of Al in excess of the Alcomplexation capacity was ca. 1.7 times greater than that in standard Al solutions. Nilsson and Powell12 observed that for many soil solutions the [Al] determined by ETAAS is lower than that determined by FIA with PCV (a result of the more rapid reaction of Al in samples than in standard solutions).FIA with stationary adsorbents The above discussion highlights the need to (i) separate the fractionation and analytical reactions in metal ‘speciation’ protocols, (ii) ensure that the medium for the analytical reaction is identical for analyte sourced in the sample or in the standard and (iii) shorten the life time of the fractionation step. These objectives can all be met by placing a stationary ‘adsorbent’ in the flow system immediately down-line from the injection valve to capture the targeted species. The ‘fractionation time’ is now limited to the contact time between the adsorbent and an element of flowing solution.The captured analyte can then be eluted, selectively, into a ‘clean’ carrier solution for down-line analysis. This paper compares a ‘stationary adsorbent’ FIA protocol with conventional FIA for determination of ‘reactive Al’ in soil solutions. It is applied to determination of ‘reactive Al’ in soil solutions derived from plant growth studies and to the determination of fulvic acid complexation capacity. Experimental The FIA manifold is shown in Fig. 2. Oxine was covalently immobilised onto a porous styrene–divinylbenzene polymer of 50–100 mm diameter13 and packed in a 20 ml column (2.0 mm id) which was constructed of polycarbonate with Omnifit (Cambridge, UK) end fittings. Full experimental details for the manifold and FIA protocol are given elsewhere.14 Immediately following sample injection (650 (ml) the ‘reactive Al’ [Al3+ + Al(OH)2+ + any Al in highly labile complexes] is captured on the column and separated from the sample matrix.The sample zone proceeds down-line and any ‘moderately labile’ Al undergoes reaction with the CAS reagent during the elapsed time in the reaction coil. The resulting signal (Fig. 3, ‘noncolumn reactive Al’) represents an operationally defined fraction. The signal is quantified against standard Al(OH)42 standards injected in 0.02 m NaOH. The captured Al is then eluted into the reagent stream with a smaller volume of 0.02 m NaOH (250 ml), effecting an approximately 10-fold preconcentration because the Al is mostly eluted in the first 50 ml of eluent.12 This decrease in sample zone volume is evident from the relative half widths of the peaks shown in Fig.3. The eluted Al(OH)42 experiences an identical reaction environment to that for eluted Al standards. The linear working range is 0.3–16 mm, the LOD 70 nm and the RSD at 2 mm Al is 3.7%. Root elongation studies were effected using lucerne (Wairau sp.). Seeds were germinated for 3 days on wet filter papers in covered petri dishes and then set out in pots of unamended soil (300 g) and loosely covered with 2–3 mm soil. Three pots with Fig. 1 FIA manifold for determination of Al by reaction with PCV. Carrier: Milli-Q water; reagent: 5.0 mm PCV (0.24 mm at the detector); buffer: 2.0 m hexamine, pH 6.0. The flow rates for the respective lines are given.All tubing 0.51 mm id microline (Cole Parmer, Niles, IL, USA); reaction coil, 100 cm knitted microline; sample volume, 250 ml. Fig. 2 Schematic diagram for the adsorption–elution flow-injection manifold. The chemical components of this system were: (i) carrier solution = 0.05 m NaOAc–0.05 m NaCl (pH 5.0), (ii) buffer = 2.0 m acetate buffer (pH 5.3; 5.00–5.05 at detector) and (iii) reagent = 2 mm CAS. The flow rates for the respective lines are given. The sample and eluent injection loop volumes were 650 ml (or 250 ml) and 250 ml, respectively. The reaction zone was a 300 cm knitted microline coil. 798 Analyst, May 1998, Vol. 123five seeds were used for each soil. Germination and plant growth were effected at 20 °C using a 12 h light–dark cycle. After 4 days all the seedlings were removed and the tap root lengths measured.Immediately following harvest, soil solutions were extracted from ca. 100 g of soil (roots and stones removed) by centrifugation (30 min at 3000 rpm). Membrane filtration to 0.025 mm was immediately effected on the filtrate and the solution analysed by FIA within 2 h. The soil solution pH was measured with a Russell (Auchtermuchty, Scotland) CMAWL/ 4/5/S7 combination microelectrode coupled to a Radiometer (Copenhagen, Denmark) PHM64 pH meter. Soil fulvic acid was isolated from International Humic Substances Society (IHSS) reference peat by the acid–pyrophosphate –XAD-7 method of Gregor and Powell.15 For Alcomplexation capacity measurements, 60 ml of fulvic acid solution (final concentration 17 mg l21) were adjusted to pH 4.7 with 0.05 m KCl and 0.005 m acetate.Standardised Al3+ (1.59 mm, pH 3) was added incrementally from a Gilmont micrometer syringe and the solution mixed for 3 min after each addition and before removal of a 1 ml aliquot into a plastic syringe. These aliquots from a titration were held for 2 h at room temperature before ‘free Al’ analysis by FIA. The aluminium binding curve was determined for the pH range 2.5–7.0 by incremental addition of KOH to a solution containing 0.05 m KNO3, 9 mM Al3+ and 17 mg l21 fulvic acid. Aliquots were removed at pH intervals and stored in plastic syringes for 2 h before analysis for ‘free Al’ by FIA. Model systems Several model systems were studied to establish that only ‘free Al’ is captured by the column and eluted by 0.02 m NaOH.The ligands tested were OH2 and fluoride, malonate, oxalate, citrate and tartrate. Measurements were made both in this flow system and by use of the column off-line in a ‘batch’ mode.16 For LNOH2, aged, hydrolysed 10 mm Al solutions were prepared in the pH range 4.6–6.0. For the organic ligands solutions containing 15 mm Al and 0–100 mm ligand were prepared and aged 24 h. Solution composition was calculated using published stability constants3 and the program SOLGASWATER.17 Results and discussion Oxine is a reagent which reacts efficiently with both Al and Fe (the most common interferent for analysis of Al in environmental samples). By using a dilute NaOH eluent the captured Al3+ can be eluted as the Al(OH)42 complex and separated from the potential interferent Fe3+ which is not eluted.18 An added attribute of this protocol is that the polymer Al13(OH)32 7+ is also retained on the column.It is not eluted by 0.02 m NaOH, but can be eluted in a second elution step with 0.2 m NaOH. The polymer does not react with CAS on the FIA timescale19 unless firstly depolymerised by stopping the flow for 120 s before the alkaline sample is moved into the reagent stream.14 The depolymerisation is facilitated by placing a smaller reaction coil down-line from the gel column and before the first merging zone. This represents the first method for fractionation of the ‘Al13’ polymer in environmental samples. Fig. 4 shows the detector (spectrophotometric) responses for the three fractions of Al: I = ‘moderately reactive Al’, II = ‘free Al’ and III = ‘polymeric Al–hydroxy’ species.Model systems Model systems were studied to establish that only ‘free Al’ is captured by the column and eluted by 0.02 m NaOH. The ligands tested were OH2 and fluoride, malonate, oxalate, citrate and tartrate. Data for L = oxalate are shown in Fig. 5, in which the solid line is the calculated concentration of ‘free Al’ and the datum points are for different flow rates. A paper by Simpson et al.14 presents further results. For each of these ligands it was established that, to a good approximation, only ‘free Al’ is Fig. 3 Typical FIA output signals for Al following the injection and elution of a soil solution or humic water sample, using the manifold in Fig. 2. The first peak is the ‘pre-elution’ peak, corresponding to ‘moderately reactive’ Al which is not captured by the column.The second peak corresponds to captured Al which is eluted with 0.02 m NaOH. Fig. 4 FIA output signals modelled for Al following the injection and elution of a humic sample which contains the polymer Al13(OH)32 7+, using the manifold in Fig. 2. I = sample injection, E1 = injection of 0.02 m NaOH eluent, E2 = injection of 0.2 m NaOH eluent, followed by a 120 s stop-flow. (I) = ‘moderately reactive’ Al, (II) = ‘free Al’, (III) = polymeric Al– hydroxide species. Fig. 5 The fraction of Al measured in solutions containing 15 mm Al and 0–100 mm oxalate, plotted as a function of ligand concentration. Data are presented for experiments with column residence times 1.3 s (*) and 2.8 s (:).The ‘fraction’, Fi, was calculated as the measured response for each solution relative to the response for a 15 mm Al standard. The curve for S{[Al3+] + [Al(OH)2+] + [Al(OH)2 +]} was calculated from the thermodynamic model for the H+–Al3+–ligand system using the computer program SOLGASWATER.17 Analyst, May 1998, Vol. 123 799captured by the oxine- derivatised gel at a flow rate corresponding to a sample–column residence time of ca. 1 s (pump speed 40 rpm). This reaction time was sufficiently short to minimise the sequestration of Al from its complexes. Aluminium complexation capacity We have not previously had a convenient method for determining the Al-complexation capacity (Al-CC) of humic waters and soil solutions. The potential of this FIA technique to determine Al-CC is illustrated by the results in Fig.6. These results are from analysis of solutions produced by incremental addition of Al3+ to a buffered (pH 4.7, 0.005 m acetate) solution of fulvic acid (17 mg l21) followed by a 2 h ageing of aliquots withdrawn at each solution stoichiometry. The data for ‘free Al’ are shown for experiments in the absence (/) and presence (:) of fulvic acid. When the total [Al] added is in excess of the Al-CC the ‘free’ Al increases in proportion to this excess. Extrapolation of these ‘free’ Al data to the x-axis, as shown, indicates a (kinetic) Al-CC20 of ca. 10 mm l21. This represents the sum of ‘nonlabile’ plus ‘moderately labile’ Al (i.e., all Al which is not captured by the column). The curve (2) represents ‘moderately labile’ Al and approaches a plateau at ca.8 mm Al. The sites binding ‘moderately labile Al’ require ca. 50 mm total Al (40 mm excess Al) to become saturated, evidence that they bind Al much less strongly. Thus the ratio of ‘moderately labile Al’ to ‘inert Al’ bound to the fulvic acid is ca. 4 : 1. When the total [Al] added is in excess of the Al-CC, the sum of the ‘free’ Al (:) and ‘moderately labile’ Al (2) gives a curve which has a limiting slope Å 1.0 (indicating minimal matrix effect on the rate of reaction) and an intercept of ca. 2 mm l21, corresponding to ‘inert Al’. We have reported similar complexation capacity titration curves for raw humic waters20 and soil solutions.21 Fig. 7 provides a comparison of Al3+ and Cu2+ binding [to the same fulvic acid (FA) sample] as a function of pH.The solution stoichiometries are not identical (see Fig. 7 caption) but it is clear that at the chosen ratios of -COOH to metal ions Al3+ is bound more strongly (at lower pH). The Cu curve was determined by potentiometry at a higher [FA];22 the curve for the Cu kinetic complexation capacity (and at the lower [FA] used for Al) would lie to higher pH, amplifying the difference between the two metals. It may be inferred from these results that Cu2+ (and other heavy metals) immobilised and accumulated in humic sediments of a lake will be released upon acidification of the water column. The presence of Al3+ at elevated concentrations in acidified water will significantly enhance this release of heavy metals from humic sediments. Column versus non-column methods The advantage of separating the analyte fractionation step from the analyte–reagent reaction is seen by comparison of the ‘reactive Al’ concentrations determined for a series of soil solutions by using the conventional FIA method with PCV reagent and the oxine-derivatised gel method (Fig.8). The PCV-reactive Al concentrations (5) are all significantly higher than those for ‘free Al’ determined by the oxine column [:, Fig. 8(a)]. They are also significantly greater than the sum of ‘free Al’ and ‘moderately labile Al’ determined by the gel column [Fig. 8(a) and (b)]. In part this is related to the very aggressive reaction by PCV, but predominantly to the much slower kinetics of the Al–PCV reaction in standard solutions Fig. 6 Complexation capacity data for titration of a soil-derived fulvic acid (17 mg l21) with Al3+ at pH 4.7 (0.005 m acetate buffer).Data correspond to (:) ‘free’ Al captured by the oxine-derivatised gel and eluted with 0.02 m NaOH, (2) ‘moderately reactive’ Al, derived from the preelution peak and (/) an Al titration in the absence of fulvic acid. Fig. 7 Metal binding curves for (/) Al3+ and (-) Cu2+ with a soilderived fulvic acid. For the Al binding curve [fulvic acid] = 17 mg l21 in 0.05 m KNO3; total [Al] = 9.0 mm. The Al curve was determined by the oxine-derivatised gel method. For the Cu binding curve [FA] = 25 mg l21 in 0.1 m KNO3, total [Cu] = 9.0 mm. The Cu curve is taken from reference 22 and was determined by Cu ion-selective electrode potentiometry. Fig. 8 Comparison of analyses of ‘reactive’ Al in soil solutions using the FIA–PCV method and the oxine-derivatised gel method. (a) ‘Reactive’ Al as a function of pH, measured by (5) the FIA–PCV and (:) the oxinederivatised gel method; (-) is the sum of ‘free’ Al and ‘moderately reactive’ Al by the oxine-derivatised gel method.(b) Correlation of reactive Al determined by FIA–PCV and by the oxine-derivatised gel method (‘free’ Al + ‘moderately reactive’ Al); y = 3.2265 x + 0.0578 and R2 = 0.9289. 800 Analyst, May 1998, Vol. 123(compared with Al–ligand solutions) against which the unknowns are calibrated.11 Aluminium toxicity to plants The oxine-derivatised gel method is finding application in studies on the Al toxicity of acidic soils.12,18 Fig. 9 shows the relative root elongation values (RRE) for 4 day old lucerne seedlings plotted (a) against pH, and (b) against ‘free Al’ determined in the centrifuged soil solutions at the end of the experiment.In Fig. 9 (b) the observed correlation coefficient of 0.88 can be compared with values of 0.79 for a plot of RRE against the sum of ‘free Al’ + ‘moderately reactive Al’ and a value of 0.59 for a plot of RRE against ‘PCV-reactive Al’.12 No measurable amount of the ‘Al13’ polymer was found in any of the soils studied.23 From the distribution diagrams for the Al– OH2 system the absence of this polymer at pH < 5.0 is anticipated. Other factors possibly acting against its detection are complexation with humic colloids (which are not retained on the gel column) or adsorption on inorganic colloids.Detection systems FeII and FeIII are captured by the derivatised gel but not removed with the 0.02 m NaOH eluent.18 In the past, an inability to effect this quantitative separation of Fe and Al in homogeneous solution without addition of reducing agents has seriously confounded attempts to develop electrochemical methods for analysis of Al3+.24–26 These amperometric and voltammetric methods probe the effect of Al on the oxidation reaction of a redox-active ligand, typically a 1,2-dihydroxyaryl molecule, such as DASA,24 4-nitrocatechol (with oxine column),25 alizarin26 or PCV.27 At pH 8–9 the binding of Al to such a ligand shifts the ligand oxidation peak anodically by ca. 200 mV (Fig. 10). By monitoring the free ligand oxidation [peak (a)] using a glassy carbon or Au electrode in a flow cell, the formation of Al complexes in the sample zone is registered at the detector as a decrease in anodic current.The use of a ligand oxidation reaction at positive potentials avoids the problem of dissolved oxygen (which confounds electrochemical measurements at cathodic potentials). Recently, screen-printed electrodes doped with alizarin have been developed for the voltammetric determination of Al in flow or batch systems.28 Column dynamics Column size affects the efficiency of analyte capture. Columns with 20–80 ml capacity have been tested in batch and flow applications. They were found to effect quantitative retention of free Al for sample flow rates of ca. 1 ml min21 onto the column. Columns with gel volumes of 16 ml or less do not effect quantitative retention of the analyte at this flow rate.At smaller flow rates (gel residence times ! 2.5 s) significant amounts of Al were sequestered from moderately labile complexes. The question arises as to whether the effective reaction time is related to the residence time in the column, or controlled by the diffusion layer thickness about the 50–100 mm gel particles. This is not easily resolved by modelling because of the microporous nature of the gel and because the average distance of any element of solution from the close-packed beads of gel is of the same order of magnitude as the calculated diffusion layer thickness. Experiments using species of known lability, and species with known dissociation rate constants, are being used to establish the effective reaction time.18 Conclusions Use of FIA to effect speciation of metals in moderately labile systems requires the separation of the speciation (fractionation) process from the analyte–reagent reaction. This can be achieved by using an adsorbent or immobilised complexing agent to effect the fractionation, followed by selective elution of the analyte and down-line reaction.If this protocol is not followed then matrix components may enhance or retard the analyte– reagent reaction and thus the analyte is over- or underestimated. Furthermore the analyte–reagent contact time in typical FIA manifolds (ca. 15–30 s) is sufficient to effect substantial sequestering of a metal analyte from its various complex species. This leads to an overestimate of the ‘free’ metal fraction.The principle of separated fractionation and reaction steps is illustrated by the use of an oxine-derivatised gel to capture ‘free’ Al from complex systems in a ca. 1 s contact time. From analysis of a pre-elution peak and selective elution of species captured on the gel, three Al fractions can be defined: ‘free’ Al, ‘moderately reactive’ Al and Al–hydroxy polymers [typified by Al13(OH)32 7+]. This oxine-derivatised gel has been used previously in macrocolumns for preconcentration of Al3+ and Fig. 9 Relative root length (RRL) measurements for growth of lucerne (Wairau) seedlings in unamended soils as a function of (a) pH and (b) [Al] in the soil solution at the completion of 4 days growth (22 °C, 12 h light– dark). Soil solutions were extracted by centrifugation, followed by 0.025 mm filtration.[Al] was determined as ‘free’ Al by the oxinederivatised gel method. RRL = 100 3 (root length at pH = X)/(maximum root length). The values of RRL are derived from the mean lengths for 15 seedlings (five seeds from each of three replicates); error bars are one s. Fig. 10 Cyclic voltammograms recorded at n = 100 mV s21 for 1 mm 4-nitrocatechol in the absence (a) and presence (b) of 0.33 mm AlIII, pH 9.4. Analyst, May 1998, Vol. 123 801Mn2+ from seawater,29,30 but the dynamics, kinetics and possible miniaturisation have not previously been exploited. References 1 Florence, T. M., Analyst, 1986, 111, 498. 2 Campanella, L., Pyrznska, K., and Trojanowicz, M., Talanta, 1996, 43, 825. 3 Pettit, L. D., and Powell, H. K.J., SC-database, Stability Constant Database, IUPAC, Oxford, Academic Software, 1997. 4 Driscoll, C. T., Int. J. Environ. Anal. Chem., 1984, 16, 267. 5 Morel, F. M. M., and Hering, J., Principles and Applications of Aquatic Chemistry, Wiley Interscience, New York, 1993, p. 374. 6 Pankow, J. F., and Morgan, J. J., Environ. Sci. Technol., 1981, 15, 1155. 7 Hawke, D. J., and Powell, H. K. J., Anal. Chim. Acta, 1994, 299, 257. 8 Simpson, S. L., Sj�oberg, S., and Powell, H. K. J., J. Chem. Soc., Dalton Trans., 1995, 1799. 9 Hawke, D. J., and Powell, H. K. J., Polyhedron, 1995, 14, 377. 10 Hawke, D. J., Powell, H. K. J., and Simpson, S. L., Anal. Chim. Acta, 1996, 319, 305. 11 Simpson, S. L., Powell, K. J., Nilsson, N. H. S., and Sj�oberg, S., Anal. Chim. Acta, 1998, 359, 329. 12 Nilsson, N. H. S., and Powell, K. J., unpublished results. 13 Landing, W. M., Haraldsson, C., and Paxeus, N., Anal. Chem., 1986, 58, 3031. 14 Simpson, S. L., Powell, K. J., and Nilsson, N. H. S., Anal. Chim. Acta, 1997, 343, 39. 15 Gregor, J. E., and Powell, H. K. J., J. Soil Sci., 1986, 37, 577. 16 Downard, A. J., Powell, K. J., Akhtar, P., and O’Sullivan, B., unpublished results. 17 Eriksson, G., Anal. Chim. Acta, 1979, 112, 375. 18 Adams, M. M., and Powell, K. J., unpublished results. 19 � Ohman, L.-O., and Powell, K. J., unpublished results. 20 Hawke, D. J., Powell, H. K. J., and Gregor, J. E., Mar. Freshwater, 1996, 47, 11. 21 Hawke, D. J., and Powell, H. K. J., Aust. J. Soil Res., 1995, 33, 611. 22 Town, R. M., and Powell, H. K. J., Anal. Chim. Acta, 1993, 279, 221. 23 Adams, M. M., Nilsson, N. H. S., and Powell, K. J., unpublished results. 24 Downard, A. J., Powell, K. J., and Money, S. D., Anal. Chim. Acta, 1997, 349, 111. 25 Downard, A. J., Lenihan, R. J., Simpson, S. L., O’Sullivan, B., and Powell, H. K. J., Anal. Chim. Acta, 1997, 345, 5. 26 Downard, A. J., Powell, H. K. J., and Xu, S., Anal. Chim. Acta, 1992, 256, 117. 27 Simpson, S. L., and Powell, K. J., unpublished results. 28 Downard, A. J., O’Sullivan, B., Akhtar, P., and Powell, K. J., unpublished results. 29 Resing, J. A., and Mottl, M. J., Anal. Chem., 1992, 64, 2682. 30 Resing, J. A., and Measures, C. I., Anal. Chem., 1994, 66, 4105. Paper 7/07293G Received October 8, 1997 Accepted February 20, 1998 802 Analyst, May 1998, Vo

ISSN:0003-2654    

DOI:10.1039/a707293g

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Two simple interface designs for capillary electrophoresis–inductively coupled plasma mass spectrometry† Vahid Majidi*a and Nancy J. Miller-Ihlib a Los Alamos National Laboratory, Chemical Science and Technology Division (CST-9), MS K484, Los Alamos, NM 87545, USA b United States Department of Agriculture, Agricultural Research Service, Beltsville Human Nutrition Research Center, Food Composition Laboratory, Beltsville, MD 20705, USA The design and implementation of two different interfaces for capillary electrophoresis–inductively coupled plasma mass spectrometry (CE–ICP-MS) are described.These interfaces will allow for on-line analysis of CE effluents with ICP-MS detection. One interface is based on a concentric tube nebulizer and the other on a standard cross-flow nebulizer. These systems were investigated in parallel and their performances, under various experimental conditions, were compared. Each interface possesses a unique set of advantages and shortcomings.Recognizing that typical sample flow rates for ICP-MS are of the order of ml min21 and that the flow rates for CE are a few nl min21, some difficulties in flow compatibility are encountered. Aspects discussed include interface considerations, flow compatibility and the influence of flow rates on the overall sensitivity. Several guidelines are provided for workers interested in implementing a CE–ICP-MS instrument for elemental speciation. The Cd detection limits in rabbit metallothionein were 2.36 and 0.21 mg ml21 for the concentric and cross-flow nebulizers, respectively. Keywords: Speciation; capillary electrophoresis; elemental analysis; elemental mass spectrometry; instrumentation Trace element analysis has played a pivotal role in pharmaceutical, clinical, biological, environmental, agricultural, petroleum and nuclear applications.During the last decade, the importance of the identification of the molecular structure and the chemical environment of a given element has become the focus of many publications.1–3 The identification of the molecular origin of an elemental signal is termed chemical speciation. The reason for this recent interest in chemical speciation is the realization of the fact that the determination of total elemental concentration is not a valid representation of the chemical, biological and toxicological activity of a given element.For example, it has been demonstrated that measurement of total iron concentration in food is not a good indicator of metabolically available iron because heme iron and non-heme iron are absorbed by different mechanisms (heme iron is absorbed as the intact metal– porphyrin complex).4 Furthermore, it has been shown that the absorption of non-heme iron from meals is about 5%, whereas the adsorption of heme iron from meals can be as much as 37%.5 As a result, for a more complete picture of dietary mineral nutrition, it is imperative that the concentrations of different species of iron be documented individually by one of the currently available methods.6,7 Different selenium species have also been found to differ greatly in their biological activities. In a study by Latshaw,8 two different forms of selenium (selenite and selenoamino acids) were incorporated into the feed material of egg producing hens.It was demonstrated that selenium, when fed in the form of selenoamino acids (natural selenium), was bound more tenaciously to the liver tissue of hens.Furthermore, it was shown that when selenite was incorporated into the feed, significantly more selenium was present in the yolk than in the white of the egg. In a follow-up study of egg proteins, Latshaw and Biggert9 found that the livitin fraction had the highest and the low density fractions had the lowest selenium concentrations. They also found that the increase in selenium in egg white proteins, after feeding selenomethionine, appeared to parallel the methionine content of the proteins.The above studies on iron and selenium are just two examples that clearly demonstrate the need for rapid and robust analytical techniques to perform chemical speciation analysis. Several approaches have been outlined for elemental speciation. A comprehensive overview of coupled techniques for speciation was presented by Lobinski.10 In general, for speciation applications, atomic emission detection is frequently used for gas chromatographic separations11,12 whereas inductively coupled plasma mass spectrometry (ICP-MS) detection is favored for liquid chromatographic separations.13-15 Another viable approach for chemical speciation is to exploit the efficiency of separation technology, such as capillary electrophoresis (CE), along with the great sensitivity of ICPMS.CE is a versatile separation technique that can be adapted to the analysis for a variety of molecules.16 For example, CE has been used successfully in the determination of metal ions and anions using ultraviolet absorption and electrochemical detection.Some of the attributes responsible for the popularity of CE as a separation technique include high separation efficiencies, small sample volume requirements (a few nanoliters), minimal buffer consumption (less 1 ml) and rapid sample throughput. ICP-MS is a powerful elemental analysis technique capable of routine analyte determinations in the ng l21 concentration range.17 A number of nebulizers and interfaces are commercially available for low flow sample introduction into an ICP.When attempting to use these systems for CE applications, however, one has to consider the parameters that influence both the separation efficiencies and detection limits. Total dead volumes, composition of buffers and flow compatibilities are among the parameters that must be optimized for a successful union of CE with an ICP-MS system. One functional interface design was implemented by Olesik et al.18 They placed the grounded cathodic end of the capillary directly in a concentric pneumatic nebulizer.The electrical connection to the capillary was made by conducting silver paint on the outside of the † Presented at The Third International Symposium on Speciation of Elements in Toxicology and in Environmental and Biological Sciences, Port Douglas, Australia, September 15–19m 1997. © US Government. The US Government retains nonexclusive, royalty-free license to publish or reproduce the published form of this contribution, or to allow others to do so, for US Government purposes.The work was performed under the auspices of the US Department of Agriculture. Analyst, May 1998, Vol. 123 (803–808) 803capillary. This system was used to separate a series of simple inorganic ions. Nonetheless, wide capillary had to be used in these studies to achieve sufficient sensitivity. Another functional design for a CE–ICP-MS interface was implemented by Lu et al.,19 who employed a modified concentric nebulizer (Meinhard, Santa Ana, CA, USA) with a make-up liquid sheath flow to connect their CE system to the ICP-MS detector.They were able to obtain good results for the separation of metalloproteins, but a large pressure differential experienced by the CE capillary had to be externally compensated. In this paper, we present data on two different interface designs that allow for on-line analysis of CE effluents with ICPMS detection.These interfaces are simple to fabricate and they can be implemented easily into current commercial ICP-MS systems. Experimental Instrumentation ICP-MS For the ICP-MS work, Perkin-Elmer SCIEX (Thornhill, ON, Canada) Elan Model 5000A and 6000 instrument were used. In all experiments, the ICP was operated at 1000 W with a coolant Ar flow rate of 15.0 l min21 and an auxiliary Ar flow rate of 0.860 l min21. The carrier flow rate was changed according to the experimental requirements.The ICP-MS data collection was initiated manually, immediately after the separation voltage was turned on. The rate of data collection for each electropherogram was set to 0.5, 1 or 2 Hz depending on the experimental requirements. For each electropherogram either one, two or three channels, corresponding to different m/z ratios, were interrogated. CE Preliminary separation procedures were optimized on an Applied Biosystems (Forest City, CA, USA) Model 270A-HT capillary electrophoresis system equipped with a UV detector.For ICP-MS experiments, however, an in-house fabricated CE instrument was used. The laboratory-made instrument was favored over the commercial instrument because of the total flexibility in the selection of column lengths and sample injection procedures. This CE system consisted either of a 50 or 110 cm long (50 mm id, 363 mm od) fused-silica capillary (Polymicro Technologies, Phoenix, AZ, USA).The separation potential was produced by a Series 230 high voltage power supply from Bertan (Hicksville, NY, USA). A new capillary was conditioned by flushing 1.0 m NaOH solution through it for 30 min. Next, the capillary was flushed with distilled, deionized water for 10 min. After running each sample, the column was regenerated using 0.1 m NaOH flush for 10 min, followed by 3 min of flushing with the run buffer solution. This conditioning and regeneration regimen insures reproducible migration times from run to run.Sample injection conditions will be discussed individually for each experiment. Reagents Several buffers were used in this study. Buffers were made by dilution of appropriate amounts of crystalline acids and their corresponding conjugate bases. The dilutions were made using appropriate amounts of 18 MW cm ultra-pure water, produced with a Milli-Q water purification system (Millipore, Bedford, MA, USA).All chemicals, including the proteins and metal salt standards, were purchased from Sigma (St. Louis, MD, USA) at the highest purity available. In all cases, when pure metal solutions were prepared, the nitrate salt of each metal was used. Results and discussion We designed two interfaces that allow for the on-line ICP-MS detection of species separated by CE. The first system is based on a concentric nebulizer design and the other is developed to function with the aid of a commercially available nebulizer. The latter design can be successfully connected to any nebulizer, but we used a cross-flow nebulizer that was supplied with the Elan ICP-MS instrument.Both designs are simple, inexpensive, robust and easy to implement. Concentric nebulizer interface Using a series of quartz capillaries, a modified concentric nebulizer similar to a design by Milstein and Caruso20 was fabricated (Fig. 1). A 50 cm long CE capillary (50 mm id, 363 mm od) was placed inside an 80 mm long buffer sheath flow capillary (530 mm id, 625 mm od).This assembly was then inserted into a 20 mm long, larger capillary (700 mm id, 850 mm od) which accommodates the Ar nebulizer gas. The 24 mm thick polyamide coating had to be thermally stripped from the first 30 mm of the buffer sheath flow capillary to allow for easier insertion of this tube into the outer-most capillary. These tubes were held in place by two P-728 PEEK tees with no additional efforts to keep them centered.All tees, fittings, PEEK ferrules and Teflon tubing were purchased from Upchurch Scientific (Oak Harbor, WA, USA). The separation capillary (CE) was held in place at the end of one tee with an F-300 PEEK fitting and a 0.4 mm Vespel ferrule (Model 100/0.4-VG1, Alltech, Deerfield, IL, USA). The buffer sheath flow capillary connections with the two tees were made using two M-110 Kel-F Minitight fittings. The nebulizer quartz capillary (Ar flow capillary) was held in place at the terminal tee with a stainlesssteel fitting and a 1.0 mm Vespel ferrule (Model 100/1.0-VG1, Alltech).For this connection, a stainless-steel fitting was chosen because it does not come in contact with any solution (i.e., there is no contamination from acid leaching) and because it has a small profile that allows for easier insertion of the nebulizer into the spray chamber. To meet the liquid flow requirements of the nebulizer, a buffer flow (sheath) was mixed with the capillary effluents.The mixing of the buffers from the two capillaries occurred at the tip of the separation capillary. The optimum position of the Fig. 1 Schematic diagram of concentric nebulizer interface for CE–ICPMS. 804 Analyst, May 1998, Vol. 123separation capillary was found to be 1 mm recessed from the tip of the buffer sheath flow capillary, generating a total dead volume of 220 nl (Fig. 1). This capillary alignment was easily accomplished using a laboratory-made, 1 mm long insertion jig. The buffer sheath flow also served as an electrical connection to the cathodic end of the separation capillary.To supply the buffer flow to the interface, a section of Teflon tubing (Model 1522, 0.76 mm id 31.6 mm od) attached to a syringe pump was connected by an F-300 PEEK fitting to the buffer tee. The cathodic electrical connection (ground) was also made through the same fitting by placing a 0.076 mm od platinum wire between the Teflon tubing and the ferrule. Owing to the small diameter of the platinum wire and the flexibility of Teflon tubing, an excellent conductive liquid tight connection was obtained.Another piece of 0.76 mm id. Teflon tubing was connected by an F-120 fitting F-120 to the Ar tee for the nebulizer gas flow. For low flow nebulizers, the large volume of conventional spray chambers may degrade the peak shapes in electropherograms. Subsequently, a low volume cyclone spray chamber with a total volume of 4 ml was used for this interface.The spray chamber consisted of 14 cm 3 6.4 mm id. Tygon tubing bent into a circular shape, with one end connected to the CE nebulizer and the other end to the ICP-MS. The connection to the ICP-MS was made through the standard electrothermal vaporizer compression connector and Teflon tubing. The combination of this nebulizer and spray chamber can operate at sheath buffer flow rates in the range 10–200 ml min21. As with any pneumatic nebulizer, the flow rate of the nebulizing gas and the analyte flow rate play an important role in the analyte transfer efficiency and the overall sensitivity. The response of the concentric nebulizer as a function of Ar flow rate was evaluated by introducing 20 ml min21 of 70 mg ml21 zinc sheath solution into the ICP-MS at different nebulizer gas flow rates [Fig. 2(a)]. The Zn analyte was introduced via the sheath flow solution in order to obtain a steady-state signal.As the Ar flow rate is increased from 0.9 to 1.1 l min21 the Zn ion intensity becomes greater. When the Ar flow rate is increased beyond 1.1 l min21 the Zn signal intensity steadily decreases. Therefore, the optimum nebulizer gas flow rate for this interface is 1.1 l min21. Fig. 2(b) illustrates the influence of the buffer sheath flow rate on the analyte (70 mg ml21 Zn solution) signal intensity at the optimized Ar flow rate of 1.1 l min21. As in Fig. 2(a), the Zn was introduced into the sheath flow solution.In this experiment, the signal intensity increases monotonically as the solution flow rate is increased. The slight overshoot of the analyte intensity at the beginning of each new solution flow rate is due to the non-uniform pressure imposed by the gear change in syringe pump at the onset of each new setting. In addition to the overall signal enhancement at higher sheath buffer flow rates, we can also observe a more stable electropherogram.This phenomenon is illustrated in Fig. 3 for a pure 0.1 mm La aqueous solution that was injected electrokinetically, for 2 s at 5 kV. The run buffer was 20 mm ammonium acetate solution (pH 7.0) and the separation voltage was set to 10 kV (generating 14 mA). The injection and migration of La were investigated at sheath buffer flow rates of 5, 10, 20 and 50 ml min21. Interestingly, regardless of the buffer flow rate, the La peak always appears at the same time. This indicates that the sheath buffer flow rate does not produce a significant backpressure at the cathodic end of the capillary. However, at lower buffer flow rates, a significant memory effect is observed in the electropherogram.At 5 ml min21, several La spikes can be seen. These spikes are due to the poor transport and mixing efficiency of the CE effluent with the buffer sheath flow, prior to the nebulization. As the flow rate of the sheath buffer is increased, the magnitude of peak tailing and number of spurious spikes are reduced.The results indicate that this nebulizer can reliably perform at flow rates beyond 200 ml min21; however, because of the low volume of the spray chamber, additional analyte spikes are observed at high flow rates ( > 100 ml min21) due to release of the analyte species from the wall of the spray chamber. Subsequently, a compromise flow rate of 50 or 100 ml min21 was often used in CE–ICP-MS experiments. After the optimization of the buffer sheath flow and the Ar nebulizer gas flow rate, two different types of samples were used to evaluate the utility of this interface.The first sample was an aqueous mixture of Mn (20 mg ml21), Ni (20 mg ml21), Co (6 mg ml21) and La (14 mg ml21). This solution was injected electrokinetically for 10 s at 8 kV. The separation was achieved at a run voltage of 8 kV in 20 mm ammonium acetate buffer (pH 7.0, 14.9 mA) using a 50 ml min21 sheath buffer flow rate. The successful separation and isotope specific detection of these analytes is illustrated in Fig. 4(a). The second sample [Fig. 4(b)] was a 1 mg ml21 solution of rabbit metallothionein. Previous Fig. 2 Concentric nebulizer performance characteristics for a 70 mg ml21 Zn sheath solution injected continuously with a syringe pump. a, 64Zn ion intensity as a function of nebulizer gas flow rate for a 20 ml min21 sheath solution flow; b, 64Zn ion intensity as a function of sheath solution flow rate at a 1.1 l min21 nebulizer gas flow rate.Fig. 3 Influence of buffer sheath flow rate on migration characteristics of 0.1 mm La solution. La solution was electrokinetically injected for 2 s at 5 kV into a 50 cm long CE column; 20 mm ammonium acetate buffer (pH 7.0) at 10 kV run potential difference was used (14 mA run current); the ion intensity of 139La was measured. Analyst, May 1998, Vol. 123 805work19 has shown that this protein has three distinct cadmium containing isoforms which can be separated by CE.Because this protein contains several metals (6.7% Cd, 0.5% Zn and trace Cu), ICP-MS can be used to follow the progress of separation. This protein was injected for 5 s at 10 kV and the separation was achieved in 50 mm TRIS buffer (pH 9.1, 4.5 mA). Using the 3s criterion, the detection limit for Cd in rabbit metallothionein was determined to be 2.36 mg ml21. Standard cross-flow nebulizer interface This extremely simple, versatile and robust interface is based on a standard cross-flow nebulizer and a commercial Scott spray chamber.However, this interface can be adapted to any nebulizer and spray chamber assembly. The nebulizer diagram is shown in Fig. 5. In this interface, the 110 cm long separation capillary was inserted through a P-728 PEEK tee and it was held in place using an F-300 fitting with a 0.4 mm Vespel ferrule (Model 100/0.4-VG1, Alltech). Similarly to the previous design, a sheath buffer flow was mixed with the CE effluent prior to nebulization.Once again, the mixing of the buffer solutions occurred at the tip of the separation capillary. The buffer sheath flow also served as an electrical connection. To supply the buffer flow to the interface, a small section of Teflon tubing (Model 1522, 0.76 mm id 3 1.6 mm od) attached to a peristaltic pump, using 0.51 mm id two-stop Tygon tubing, was connected by an F-300 to the buffer tee. The cathodic electrical connection was made through the same fitting by placing a 0.076 mm od platinum wire between the Teflon tubing and the ferrule.Another piece of 0.76 mm id Teflon tubing was connected by an F-120 fitting which directed the buffer sheath flow and the separation capillary into the tip of the nebulizer. The interface was coupled to the inlet of the cross-flow nebulizer by 5 mm 3 0.76 mm id, Tygon tubing. To make this coupling, the Teflon tubing was cut at a shallow angle (to make a sharp tip) and one half of the tubing was placed in methanol (allowing it to swell) for about 30 s.The sharp end of the Teflon tube was dipped in methanol and then it was forced into the distended Tygon tubing. After the evaporation of methanol a strong, leak free coupling was obtained. The other end of the Tygon coupling can be easily slipped over the nebulizer Teflon tube. For efficient operation of this interface, a buffer sheath flow rate of 100–1000 ml min21 is suggested. The additional length of the capillary for this CE interface, does not influence any of the transport properties; however, it does influence the overall resolution.As such, we did not feel it was prudent to compare the separation resolution for the two systems. A number of workers have attempted to use the direct coupling of a CE capillary with a standard commercial nebulizer and spray chamber. However, in most instances the performance of these interfaces has been handicapped because of the self-aspirating nature of the commercial nebulizers.The aspiration rate of each nebulizer is strongly influenced by the Ar gas flow rate and the nebulizer design. This behavior and its effect on the migration rate for rabbit metallothionein in 50 mm TRIS buffer (pH 9.1) are demonstrated in Fig. 6. The bars in this plot indicate the mean migration time for a 10 s (3 kV) injection of rabbit metallothionein as a function of nebulizer gas flow rate (the Ar flow rate during the sample injection period was set at 0.7 l min21).The data indicate that as the gas flow rate increases, the mean migration velocity becomes faster. This suction of the analyte in the capillary is caused by the negative pressure at the inlet of the nebulizer. Comparing the results for an Ar flow rate of 1.1 l min21 at a 20 kV run voltage with the migration time obtained when no voltage was applied, we can Fig. 4 CE–ICP-MS electropherogram obtained with a run potential diffrence of 10 kV on a 50 cm long CE column.a, Separation of a mixed Mn (20 mg ml21), Ni (20 mg ml21), Co (6 mg ml21), and La (14 mg ml21) solution. The solution mixture was injected electrokinetically for 10 s, at 8 kV; a 20 mm ammonium acetate buffer (pH 7.0) at a 10 kV run potential difference was used (14.9 mA run current); the ion intensities of 55Mn, 58Ni, 59Co, and 139La were measured; b, Separation of cadmium containing isoforms in a solution of rabbit metallothionein (1 mg ml21).The protein solution was injected electrokinetically for 5 s, at 10 kV; a 50 mm TRIS buffer (pH 9.1) at a 10 kV run potential difference was used (4.5 mA run current); the ion intensity of 112Cd was measured. Fig. 5 Schematic diagram of a standard cross-flow nebulizer interface for CE–ICP-MS. Fig. 6 Mean migration time of rabbit metallothionein as a function of Ar flow rate in a cross-flow nebulizer interface. The protein solution was injected electrokinetically for 10 s at 3 kV into a 110 cm long CE column; a 50 mm TRIS buffer (pH 9.1) at a 20 kV run potential difference was used (4 mA run current); the ion intensity of 112Cd was measured. 806 Analyst, May 1998, Vol. 123determine that at an Ar flow rate 1.1 l min21 in a 110 cm long capillary (50 mm id), a laminar flow rate of 144 nl min21 is induced. When the linear velocity of the laminar flow rate (7.33 cm min21) is compared with the combined laminar flow, electroosmotic and electrophoretic induced migration of the analyte (47.14 cm min21), we can determine that the laminar flow rate accounts for about 16% of analyte linear velocity.This laminar portion of the flow velocity is sufficient to degrade completely the resolution of an electropherogram. To minimize the self-aspiration of the nebulizer, a series of pressure measurements were performed at the inlet of the nebulizer with the use of a water filled manometer. The water height in the manometer was allowed to equilibrate prior to the ignition of the plasma.When the plasma was ignited, the Ar nebulizer flow rate was adjusted to the desired value and the difference in the water height was recorded. These readings were then converted into a pressure differential at the inlet of the nebulizer using standard equations (corrected for the temperature). The results for these experiments are illustrated in Fig. 7, which shows that for the Scott spray chamber and crossflow nebulizer supplied with the Elan ICP-MS, there is an inherent positive pressure at the inlet of the nebulizer at nebulizer gas flow rates < 0.7 l min21.This is due to the backpressure generated by the high velocity of the plasma gas. At 0.7 l min21 there is almost no pressure differential at the inlet of the nebulizer; and at flow rates > 0.7 l min21 a negative pressure (suction) is observed at the inlet of the nebulizer which is the cause of self-aspiration. Therefore, for a CE to work properly with a commercial nebulizer and spray chamber, the nebulizer gas flow rate must be adjusted so that there is a minimum pressure differential applied to the end of the separation capillary.Along with the nebulizer gas flow rate, the rate of the sheath buffer flow must also be optimized. The flow velocity of the sheath buffer flow can be adjusted to compensate partially for the self-aspiration effect of the nebulizer. This is illustrated in Fig. 8, where the nebulizer gas flow rate was intentionally set at 0.8 l min21 to impose a slight negative pressure on the CE capillary.In these experiments, at different buffer sheath flow rates, rabbit metallothionein was injected for 3 s at 3 kV using a nebulizer flow rate of 0.7 l min21 (i.e., no analyte suction into the CE column during electrokinetic injection). Then at the onset of the CE run (25 kV in 50 mm TRIS buffer) the nebulizer gas flow rate was set to 0.8 l min21 and the ICP-MS data were collected as a function of migration time.From Fig. 8 we can conclude that for a 0.8 l min21 nebulizer gas flow, sheath buffer flow rates of < 0.375 ml min21 cannot compensate for the negative pressure of the nebulizer (imposed on the CE capillary). However, at buffer sheath flow rates of > 0.375 ml min21 the influence of nebulizer suction is completely eliminated. The advantage of using a slightly higher nebulizer gas flow rate is the gain in nebulization efficiency.Interestingly, the negative pressure at the inlet of the nebulizer can be used advantageously. This suction can be used to force conditioning solutions and cleaning buffers through the capillary, which greatly enhances the flexibility of a CE–ICPMS system. Furthermore, using a predetermined pressure change with specific duration we can inject precise amounts of sample solution onto the CE column. Fig. 9(a) depicts the migration profiles for rabbit metallothionein samples injected using a nebulizer gas flow rate of 1.0 l min21 for 5, 10 and 20 s.The nebulizer gas flow rate was then lowered to 0.7 l min21 (no suction) and the CE run was initiated with a buffer sheath flow rate of 0.375 ml min21 (separation voltage 25 kV in 50 mm TRIS buffer). Fig. 9(b) illustrates the height of the most intense metallothionein fraction as a function of injection duration. Fig. 9 clearly demonstrates that nebulizer gas flow can be used effectively for quantitative sample injection onto the CE column (r2 = 0.91).Lastly, the sensitivity of the instrument using this interface can be best illustrated with detection limits obtained for other metals in rabbit metallothionein. To illustrate this point, a solution of 1 mg ml21 rabbit metallothionein was injected at a Fig. 7 Pressure differential observed at the inlet of the cross-flow nebulizer as a function of nebulizer gas flow rate. Fig. 8 Influence of the sheath buffer flow rate on migration rate of rabbit metallothionein.The protein solution was electrokinetically injected for 3 s, at 3 kV into a 110 cm long CE column; a 50 mm TRIS buffer (pH 9.1) at 25 kV run potential difference was used (5 mA run current); the ion intensity of 112Cd was measured. Fig. 9 CE–ICP-MS electropherogram for rabbit metallothionein. a, Injection of analyte using a 1.0 l min21 nebulizer gas flow rate with different injection durations; b, Signal intensity observed for the most abundant Cd containing metallothionein fraction as a function of injection duration.The protein solution (1 mg ml21) was loaded into a 110 cm long CE column by pressure-induced injection; a 50 mm TRIS buffer (pH 9.1) at a 25 kV run potential difference was used (5 mA run current); the ion intensity of 112Cd was measured. Analyst, May 1998, Vol. 123 8071.0 l min21 nebulizer gas flow rate for 5 s. The nebulizer gas flow rate was then lowered to 0.8 l min21 and the CE run was initiated with a buffer sheath flow of 0.375 ml min21.Using a TRIS buffer at a 25 kV running voltage, the ion intensities for 112Cd, 63Cu and 64Zn were recorded as a function of migration time. The results are shown in Fig. 10. These profiles demonstrate that different portions of metallothionein proteins have different Cd : Zn : Cu ratios. Using the 3s criterion, the detection limit for Cd in rabbit metallothionein was determined to be 0.21 mg ml21. Conclusion The concentric nebulizer assembly is relatively easy to fabricate using off-the-shelf materials.This interface can perform well under a variety of conditions; however, care must be taken during the assembly of the tubes to obtain a consistent distance between the three concentric tubes. Critical parameters that must be considered prior to fabrication of this interface include the relative position of the concentric tubes, the absolute id and od of the concentric tubes, the volume of the spray chamber, the flow rates of the nebulizer gas and the sheath buffer solution, the surface tension of the buffer solution and the interface material.The judicious use of nebulizer gas and the buffer sheath flow rates makes the CE interface with a standard cross-flow nebulizer an extremely powerful, sensitive and flexible tool. Using the inherent capabilities of the nebulizer (gas flow rate), we can fabricate a useful system for cleaning and conditioning of the CE capillary and for accurate and reproducible sample injection.This interface is robust and provides a stable environment from day to day. Initially, the first interface operated for 6 months without the need for additional CE columns or cleaning. We had to change the CE column only after the capillary was accidentally broken during a routine ICPMS maintenance procedure. The evaluation of detection limits for Cd in rabbit metallothionein reveals that a factor of 10 improvement is observed for the standard cross-flow nebulizer as compared with the concentric tube nebulizer. The authors thank the Perkin-Elmer Corporation for generously providing the Elan 5000 spectrometer and the Applied Biosystems capillary electrophoresis system used in this research.The authors are grateful for the valuable assistance provided by Mrs. F. Ella Green. This work was performed in the Food Composition Laboratory of the US Department of Agriculture located in the Beltsville Human Nutrition Research Center, Beltsville, MD.The funding for this work was provided by the US Department of Agriculture, with primary funding for one author (V.M.) coming through a Cooperative Research and Development Agreement. Additonal funding for this author (V.M.) to prepare the final manuscript was provided through the TIA programme by Los Alamos National Laboratory. Mention of trademark or proprietary products does not constitute a guarantee or warranty of the product by the US Department of Agriculture or Los Alamos National Laboratory and does not imply their approval to the exclusion of other products that may be suitable.References 1 Trace Metal Analysis and Speciation, ed. Krull, I. S., Elsevier, New York, 1991. 2 Metal Speciation and Contamination of Soil, ed. Allen, H. E., Huang, C. P., Bailey, G. W., and Bowers, A. R., Lewis, Chelsea, MI, 1994. 3 Chemical Speciation in the Environment, ed. Ure, A. M., and Davidson, C. M., Blackie, Glasgow, 1995. 4 Bothwell, T. H., Charlton, R. W., Cook, J. D., and Finch, C. A., Iron Metabolism in Man, Blackwell, Oxford, 1979. 5 Hussain, R., Walker, R. B., Layrisse, M., Clark, P., and Finch, C. A., Am. J. Clin. Nutr., 1965, 16, 464. 6 Schricker, B. R., Miller, D. D., and Stouffer, J. R., J. Food Sci., 1982, 47, 740. 7 Smith, C. M. M., and Harnly, J. M., J. Anal. At. Spectrom., 1997, 12, 1055. 8 Latshaw, J. D., J. Nutr., 1975, 105, 32. 9 Latshaw, J. D., and Biggert, M. D., Poult. Sci., 1981, 60, 1309. 10 Lobinski, R., Appl. Spectrosc., 1997, 51, 260A. 11 Bulska, E., J. Anal. At. Spectrom., 1992, 7, 201. 12 Lobinski, R., and Adams, F. C., Trends Anal. Chem., 1993, 12, 41. 13 Vela, N. P., Olson, L. K., and Caruso, J. A., Anal. Chem., 1993, 65, 585A. 14 Quijano, M. A., Gutiérrez, A. M., Pérez-Conde, M. C., and Cámara, C., J. Anal. At. Spectrom., 1996, 11, 407. 15 Ødegård, K. E., and Lund, J. W., Anal. At. Spectrom., 1997, 12, 403. 16 Baker, D. R., Capillary Electrophoresis, Wiley, New York 1995. 17 Inductively Coupled Plasmas in Atomic Spectrometry, ed. Montaser, A., and Golightly, D. W., VCH, New York, 1992. 18 Olesik, J. W., Kinzer, J. A., and Olesik, S. V., Anal. Chem., 1997, 67, 1. 19 Lu, Q., Bird, S. M., and Barnes, R. M., Anal. Chem., 67, 2949. 20 Milstein, L. S., and Caruso, J. A., paper presented at the Pittsburgh Conference on Analytical Chemistry and Applied Spectrometry, Atlanta, GA, March 1997, paper 597. Paper 7/07770J Received October 28, 1997 Accepted February 27, 1998 Fig. 10 Detection of Cd, Cu and Zn in specific isoforms of rabbit metallothionein (1 mg ml21) with the use of CE–ICP-MS. The protein solution (1 mg ml21) was loaded for 5 s into a 110 cm long CE column using pressure-induced injection with a nebulizer gas flow rate of 1.0 l min21; a 50 mm TRIS run buffer (pH 9.1) at a 25 kV run potential difference was used (5 mA run current); the ion intensity of 112Cd was measured. 808 Analyst, May 1998, Vol. 123

ISSN:0003-2654    

DOI:10.1039/a707770j

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Potential sources of error in capillary electrophoresis–inductively coupled plasma mass spectrometry for chemical speciation† Vahid Majidi*a and Nancy J. Miller-Ihlib a Los Alamos National Laboratory, Chemical Science and Technology Division (CST-9), MS K484, Los Alamos, NM 87545, USA b United States Department of Agriculture, Agriculture Research Service, Beltsville Human Nutrition Research Center, Food Composition Laboratory, Beltsville, MD 20705, USA The distribution concentration of chemical species in a sample is dictated by the physical and chemical properties of the matrix.As such, when a sample is pre-treated, in any way, there is a potential for redistribution of homologous species. The extent of this analyte redistribution is determined by both thermodynamic properties of species (e.g., changes in concentrations of species according to their equilibrium expressions) and kinetic properties (e.g., the rate of the reactions compared with the duration of sample preparation and analysis).The redistributions of analyte species as a function of several experimental parameters (e.g., time, solution pH, injection methods and calibration methods) are illustrated in this paper. Whereas rabbit metallothionein protein showed a stability of more than a few days under certain storage conditions, coenzyme-B12 was rapidly degraded in less than 2 h. pH studies showed that the migration of free Cd2+ ions in rabbit metallothionein was not significantly affected unless the pH of the solution exceeds the solubility limit of the metal hydroxide. However, pH-sensitive compounds such as vitamin B12 showed significant changes in the migration time and analyte composition.The injection studies suggested that electrokinetic injection may produce biased results, in favor of species that have higher electrophoretic mobility. Hydrodynamic injection will produce a result that is more representative of the initial sample composition.Keywords: Speciation; capillary electrophoresis; elemental mass spectrometry; cyanocobalamin; rabbit metallothionein; coenzyme-B12 The rapid expansion of the chemical speciation field is due, in part, to the development of sensitive and selective coupled analytical techniques. Using modern separation procedures along with element specific detection, it is now possible to evaluate quantitatively the different forms of a specific element in a relatively complex matrix.1 However, for these techniques to become universally accepted, one must demonstrate measurement traceability.2 Traceability is an unbroken chain of calibration events that must connect the measurement process to the fundamental units; for elemental speciation, it must also be verified that the chemical species have been preserved.The errors due to coupling of separation techniques with element-specific detectors2 and the changes in species concentration as a function of thermodynamic stability of a sample3,4 and kinetic stabilities of a sample for different techniques5 have been discussed previously.Often, the new literature is based on the development of a new approach to speciation.6–8 Most of these techniques have been tested with stable chemical species within the optimun range of instrumental parameters that provide the best detection limit or signal-to-noise ratio for a given analyte (or class of compounds) in a specific matrix.The majority of the remaining new publications on speciation are focused on using a given technique for a specific analyte (or class of compounds) in a defined matrix.9–12 As the analyte of interest, compound stability or the sample matrix is changed for a given published technique, the resulting data may become suspect. Each technique may have a limitation that is specific to a given sample or an instrumental parameter. For example, Olesik et al.13 used ionspray mass spectrometry to evaluate the speciation of the Ni–EDTA complex.Although they were not able to observe the Ni–EDTA complex, their results indicated that the Ni2+ signal correlated with the expected free metal ion concentration in the solution (not the total metal concentration). Therefore, for this particular instrumental approach, if the analyte matrix is not known (this is the case with all real samples), the result, although reproducible, is misleading. The authors acknowledged this shortcoming by stating that the ‘...errors in quantitative analysis by electrospray or ionspray mass spectrometry may be severe...’.If a technique for speciation involves chemical reactions, the rate or the lack of uniform reactivity towards all similar species can be troublesome. In a few instances, this limitation can be used advantageously for screening or first-order speciation. Willie14 used this approach for the speciation of As species that react with sodium tetrahydroborate.Although individual species could not be identified, using hydride generation atomic absorption spectrometry Willie was able to evaluate the combined concentrations of AsIII, AsV, monomethylarsonic acid and dimethylarsinic acid. The limitations of conventional chromatographic separation combined with element-specific detection for speciation have been described by Quevauviller.2 Capillary electrophoretic (CE) techniques have many advantages over conventional separation technologies. The lack of a stationary phase (better representation of labile analytes due to the absence of extemporaneous interactions with the materials in the stationary phase), minimal sample and buffer (mobile phase) volume requirements, ability to separate cationic, anionic and neutral species (with the use of modifiers) and high separation efficiencies are some of the attributes that make CE a potentially indispensable tool for elemental speciation.Recent advances and a better understanding of nebulizers used in inductively coupled plasma mass spectrometry (ICP-MS) have led to a series of functional interface designs for CE–ICP-MS applications. 6,15,16 Interestingly, the use of CE separation with ICP-MS † Presented at The Third International Symposium on Speciation of Elements in Toxicology and in Environmental and Biological Sciences, Port Douglas, Australia, September 15–19, 1997. © US Government. The US Government retains nonexclusive, royalty-free license to publish or reproduce the published form of this contribution, or to allow others to do so, for US Government purposes. The work was performed under the auspices of the US Department of Agriculture. Analyst, May 1998, Vol. 123 (809–813) 809detection for speciation has brought to light several parameters that may contribute to misleading results. Because fused-silica columns used in CE have small internal diameters (25–100 mm), real samples have to be pre-treated to some extent to enhance sensitivity, facilitate separation or prevent physical damage to columns (e.g., clogging). This pre-treatment can be relatively simple (e.g., filtration, adjustment of pH or ionic strength, extraction, preconcentration and fractionation) or extremely complex (e.g., derivatization, chemical labeling, enzyme digestion and denaturation).In this paper, a series of experiments are presented that illustrate the above principles as they relate to chemical speciation.Sampling bias, chemical stability and temporal degradation of chemical species were investigated using CE– ICP-MS. Experimental Instrumentation ICP-MS A Perkin-Elmer SCIEX (Thornhill, ON, Canada) Elan 6000 instrument was used for the ICP-MS work. In all experiments, the ICP was operated at 1000 W with an Ar plasma flow rate of 15.0 l min21 and an auxiliary Ar flow rate of 0.860 l min21. The carrier flow rate (nominally 0.7 l min21) was changed according to the experimental requirements.6 The ICP-MS data collection was initiated manually, immediately after the separation voltage was turned on.The rate of data collection for each electropherogram was set to 0.5, 1 or 2 Hz depending on the experimental requirements. For each electropherogram either one, two or three channels, corresponding to different m/z ratios, were interrogated (112Cd, 114Cd, 63Cu, 65Cu, 64Zn, 66Zn and 59Co). CE For the CE–ICP-MS interface, a laboratory-made instrument was favored over the commercial instrument because of the total flexibility in the selection of column lengths and sample injection methodologies. This interface is described in detail elsewhere.6 Briefly, the laboratory-made CE system consisted of a 110 cm long (50 mm id, 363 mm od) fused-silica capillary (Polymicro Technologies, Phoenix, AZ, USA).The separation potential was produced by a Series 230 high voltage power supply from Bertan (Hicksville, NY, USA). A new capillary was conditioned by flushing 1.0 m NaOH solution through it for 30 min.Next, the basic solution was removed from the capillary by flushing the column with distilled, de-ionized water for 10 min. After running each sample, the column was regenerated by a 0.1 m NaOH flush for 10 min, followed by 3 min of flushing with the run buffer solution. This conditioning and regeneration regimen insures reproducible migration times from run to run. Reagents TRIS buffers were made by dilution of appropriate amounts of crystalline acids and their corresponding conjugate bases.The dilutions were made using appropriate amounts of 18 M½ cm ultra-pure water, produced with a Milli-Q water purification system (Millipore, Bedford, MA, USA). All chemicals, including the proteins and metal salt standards, were purchased from Sigma (St. Louis, MO, USA) at the highest purity available. Results and discussion Owing to the lack of a stationary phase, CE is the least invasive separation technique for chemical speciation.Stationary phases in various chromatographic techniques can interact with labile species and ultimately disturb the original distribution of species. The analytical results on chemical speciation can be significantly biased depending on how the experiment was conducted, how long the sample has been stored, whether the sample was pre-treated and what type of separation procedure was employed. The following discussion focuses on a few of the most common factors that can give rise to erroneous results for chemical speciation. We should emphasize that most of the potential errors highlighted in this paper are not unique to CE techniques.Sample storage and aging Sample aging begins when the sample is removed from its native environment. In biological systems, organisms actively function to maintain a balance between various chemical components (chemical species). Once the sample is removed from its native environment, kinetic and thermodynamic factors will dictate the new distribution of species.This is also true for environmental samples. Fig. 1 illustrates the effect of aging on a rabbit metallothionein (MT) solution. Metallothionein is a cysteine rich protein with an average molecular mass of 6–7 kDa. These proteins are believed to be synthesized as a direct response to metal (Cd, Cu, Zn, Ag and Hg) exposure by the biological organisms. A typical 1 mg ml21 solution of rabbit Cd-MT solution will generate two major peaks when separated by CE (with UV absorption detection) in a 50 mm TRIS buffer (pH 9.1).17 The CE–ICP-MS electropherogram (Cd ion intensity) of fresh 1 mg ml21 rabbit Cd-MT, injected electrokinetically at 10 kV for 5 s and separated at 30 kV, as a function of migration time is shown in Fig. 1(a). As expected, two major Cd peaks are observed for this sample, corresponding to the different isoforms of the rabbit MT. When this sample was allowed to sit for 2 weeks without refrigeration, the CE–ICPMS electropherogram in Fig. 1(b) was obtained. Metallothioneins have a tendency to degrade under unfavorable conditions, and as is apparent from the lack of well defined individual components in Fig. 1(b), this sample has degraded. It is important to note that the lack of separation resolution is not due to poor adjustment of the CE conditions but rather a result of sample instability. Another example of sample aging can be seen in Fig. 2, which illustrates a time-dependent Co ion intensity electropherogram for coenzyme-B12, generated by CE–ICP-MS, as a function of migration time. In this case, a pure solid sample of coenzyme- B12 was removed from the refrigerator and dissolved in a deoxygenated buffer to provide a 1 mg ml21 solution. However, the sample solution was allowed to sit uncapped, in an ambient Fig. 1 Cd ion intensity in CE–ICP-MS of rabbit metallothionein in TRIS buffer (pH 9.1). a, Fresh preparation of metallothionein solution; b, an unrefrigerated, 2 week old preparation of rabbit metallothionein. 810 Analyst, May 1998, Vol. 123environment, and a CE–ICP-MS profile was obtained for this solution about every 30 min. As the oxygen dissolved in the buffer solution, the coenzyme-B12 began to oxidize. Two peaks were observed in the CE–ICP-MS electropherogram only 10 min after the solution was prepared. The peak appearing later (approximately 17 min) was due to coenzyme-B12 and the peak appearing earlier (approximately 15 min) was indicative of the oxidation product of coenzyme-B12.The broad shoulder immediately after the oxidized coenzyme-B12 is due to incomplete separation of the oxidation products during the migration (i.e., coenzyme-B12 continues to be oxidized in the separation capillary). The intensity of the oxidized coenzyme- B12 becomes nearly three times as large as that of the unoxidized coenzyme at 43 min after the solution is prepared, and by 108 min after the preparation of the coenzyme solution the majority of the cobalt species are in the form of an oxidized analog of coenzyme-B12.Sample injection parameters Sample injection is seemingly one of the simplest operations for elemental speciation; however, in CE techniques, the type of injection and the injection duration are two critical parameters that may bias the results and lead to poor separation and irreproducible migration times.Because the inner diameter of the CE column is very small (i.e., low tolerance for sample loading), longer injection durations can severely degrade the resolution of electropherograms. This column overloading is demonstrated in Fig. 3 for both electrokinetic (sample injection achieved using a potential difference) and hydrodynamic (sample injection achieved using a pressure difference) injections. In these experiments, rabbit Cd-MT was injected on to the column and separated in 50 mm TRIS buffer (pH 9.1) at a 25 kV running potential.Different durations of hydrodynamic injection [Fig. 3(a)] were performed by suction (6200 Pa) at the cathodic end of the capillary. The electrokinetic injections [Fig. 3(b)] were done by imposing a 3 kV potential difference between the sample (anode) and the cathodic end of the capillary. In Fig. 3(a) (hydrodynamic injection), it can be seen that with shorter sample injection durations (e.g., 5 s injection), near baseline resolution is obtained for different cadmium containing components.As the injection duration becomes longer, the column becomes overloaded and the resolution is degraded to the point that for a 60 s injection duration it is difficult to discern the number of components present. When using electrokinetic injection, the separation resolution is preserved for longer injection durations because during the injection period the sample is actually undergoing separation. As such, the band overlap is not as severe as it is for hydrodynamic injection.Nonetheless, because of this preseparation, during the injection period, the migration times for a given component with different injection periods will not be the same. From the above experiments, one may conclude that electrokinetic injection is superior to hydrodynamic injection for chemical speciation. However, because electrokinetic injection uses a potential difference for sampling, the analytes with higher electrophoretic mobility will be preferentially loaded on to the column. Hence, depending on the difference between the electrophoretic mobilities of the components, substantial sampling bias may be introduced in the analysis. This concept is shown in Fig. 4, where the peak heights for two of the cadmium containing isoforms are illustrated. The hydrodynamic injections in these experiments were performed using a 3200 Pa pressure difference and the electrokinetic injections using a 3 kV injection potential.In Fig. 4(a), it can be Fig. 2 Time dependent oxidation of coenzyme-B12 measured by CE–ICPMS. Fig. 3 Influence of injection duration on separation efficiency and resolution in CE–ICP-MS. a, Hydrodynamic injection at 6200 Pa; b, electrokinetic injection at 3 kV. Fig. 4 Influence of injection parameters on sampling accuracy in CE–ICPMS. Solid lines represent the Cd ion intensity from the largest peak in Cd- MT and the dashed lines represent the Cd ion intensity from the second largest peak in Cd-MT.a, Hydrodynamic injection at 3200 Pa; b, electrokinetic injection at 3 kV. Analyst, May 1998, Vol. 123 811seen that the intensity of the cadmium peak for the two different isoforms increases linearly as a function of the injection period. Furthermore, the ratio of these two components remains nearly constant for all injection periods. The peak intensity of the cadmium containing components is not a linear function of the injection duration and the ratio of the two peaks varies significantly for electrokinetic injection [Fig. 4(b)]. In general, electrokinetic injection can provide better resolution for longer injection periods and can lead to the preconcentration of certain analytes (better detection limits). However, these improvements are at the expense of sampling accuracy. By using electrokinetic injection, severe sampling bias will result in analytical data which are counter to the fundamental principle of speciation.Influence of pH pH plays an important role in CE separations and in the chemical distribution of pH sensitive species. For example, by adjusting the pH of a biological sample to a slightly basic value, the analyst can greatly simplify the separation and identification of different isoforms of cadmium metallothionein. However, any relevant information concerning free metal and weakligand complexes of cadmium is lost owing to formation of hydroxide precipitates; Ksp for Cd(OH)2 is 5.9 310215.The Cd ion intensities obtained from the CE–ICP-MS electropherogram for a 1 mm cadmium chloride solution at different pHs are shown in Fig. 5(a). According to the solubility constant for Cd(OH)2, at pH values less than 8, more than 6 mm of free Cd can be present in the solution without significant hydroxide formation. Subsequently, for CE–ICP-MS runs at pHs 4.5, 6.0 and 7.0 we see Cd ion intensities of similar magnitude. The reason for the earlier appearance of the Cd peak at increasing pH values is that the electroosmotic flow becomes more pronounced as the pH increases.At pH 8.5, the Cd ion intensity becomes much smaller and broader than in previous runs. At this pH, according to the solubility constant, only 0.59 mm of Cd can exist as free ions. At pH 9.0 only 59 mm of free Cd can be present in the solution (at pH 9.0 the Cd signal generated by CE–ICP-MS was small, but detectable). The influence of pH becomes more problematic if the analytes present in the sample can inter-convert into one another.This is demonstrated by analyzing a pharmaceutical preparation of vitamin B12 by CE–ICP-MS. Szpunar18 has shown that the pharmaceutical preparation of vitamin B12 is composed of three cobalt containing components. Furthermore, it is known that vitamin B12 is most stable in the pH range 4.5–5.19 The Co ion intensities obtained from the CE–ICP-MS electropherogram for a 1 mg ml21 solution of vitamin B12 at different pHs are shown in Fig. 5(b). At pH 4.5, as reported by Szpunar,18 three cobalt containing peaks are observed for the vitamin B12 solution. These peaks are assigned to free cobalt (early peak), hydroxycobalamin and adenosylcobalamin. At pH 6.0, these three peaks are shifted towards later times. This shift is most likely due to deprotonation of the phosphate moiety on the parent ring system (i.e., changing the electrophoretic mobility) and coordination of free cobalt with some hydroxides.At pH 7.0, the electroosmotic flow becomes strong enough to early shift the peaks, while the central band is lost. At pH 9.0 it is likely that the Co is removed from the porphyrin structure in order to form Co(OH)2. Conclusion The combination of CE with ICP-MS is an extremely valuable tool for elemental speciation. As with any sophisticated instrumental technique, the users must be aware of the possible analytical difficulties that may be associated with sample handling and/or each of the individual techniques (CE and ICPMS) in addition to the combined technique (CE–ICP-MS).Interestingly, none of the potential sources of error are unique to CE–ICP-MS. A good protocol for sample preparation, storage and calibration is needed to avoid sample aging and facilitate quantitative analysis. The pH influence on the analyte can be minimized by running CE with a pH buffer that is similar to that of the sample. Fortunately, CE can operate over a wide pH range (2–11).Column overloading is also a common problem with all separation techniques. Employing smaller sample volumes and using peak shapes as a guide, overloading of the column can be avoided. The only potential source of error that was unique to CE is the sampling bias observed during electrokinetic injection. This bias can be easily eliminated if the hydrodynamic injection technique is used for CE–ICP-MS analysis. With careful preparation of samples, and judicious selection of sampling techniques and analysis conditions, CE–ICP-MS can become a benchmark technique for elemental speciation.The authors are grateful for the valuable assistance provided by Mrs. F. Ella Green. This work was performed in the Food Composition Laboratory of the US Department of Agriculture located in the Beltsville Human Nutrition Research Center, Beltsville, MD. The funding for this work was provided by the US Department of Agriculture, with primary funding for one author (V.M.) coming through a Cooperative Research and Development Agreement.Additional funding for this author (V.M.) to prepare the final manuscript was provided through the TIA program by Los Alamos National Laboratory. Mention of trademark or proprietary products does not constitute a guarantee or warranty of the product by the US Department of Agriculture or Los Alamos National Laboratory and does not imply their approval to the exclusion of other products that may be suitable.References 1 Lobinski, R., Appl. Spectrosc., 1997, 51, 260A. 2 Quevauviller, P., J. Anal. At. Spectrom., 1996, 11, 1225. 3 Stewart, I. I., and Horlick, G., J. Anal. At. Spectrom., 1996, 11, 1203. 4 Donat, J., Lao, K. A., and Bruland, K. W., Anal. Chim. Acta, 1994, 284, 547. 5 Lu, J. Y., Chakrabarti, C. L., Back, M. H., Sekaly, A. L. R., Gregoire, D. C., and Schroeder, W. H., J. Anal. At. Spectrom., 1996, 11, 1189. Fig. 5 Influence of pH on migration time and distribution of species. a, Effect of pH on free Cd; b, effect of pH on a pharmaceutical preparation of vitamin B12. 812 Analyst, May 1998, Vol. 1236 Majidi, V., and Miller-Ihli, N. J., Analyst, 1998, 123, 803. 7 Wang, L., May, S. W., Browner, R. F., and Pollock, S. H., J. Anal. At. Spectrom., 1996, 11, 1137. 8 Corr, J. J., and Anacleto, J. F., Anal. Chem., 1996, 68, 2155. 9 Pergantis, S. A., Winnik, W., and Betowski, D., J. Anal. At. Spectrom., 1997, 12, 531. 10 Cuesta, A., Todoli, J. L., and Canals, A., Spectrochim. Acta, Part B, 1996, 51, 1791. 11 Crews, H. M., Clarke, P. A., Lewis, D. J., Owen, L. M., Strutt, P. R., and Izquierdo, A., J. Anal. At. Spectrom., 1996, 11, 1177. 12 Ding, H., Olson, L. K., and Caruso, J. A., Spectrochim. Acta, Part B, 1996, 51, 1801. 13 Olesik, J. W., Thaxton, K. K., and Olesik, S. V., J. Anal. At. Spectrom., 1997, 12, 507. 14 Willie, S. N., Spectrochim. Acta, Part B, 1996, 51, 1781. 15 Olesik, J. W., Kinzer, J. A., and Olesik, S. V., Anal. Chem., 1995, 67, 1. 16 Lu, Q., Bird, S. M., and Barnes, R. M., Anal. Chem., 1995, 67, 2949. 17 Richards, M. P., Beattie, J. H., and Self, R., J. Liq. Chromatogr., 1993, 16, 2113. 18 Szpunar, J., paper presented at the 43rd International Conference on Analytical Sciences and Spectroscopy, August 1997, Montréal, paper 3-5. 19 The Merck Index, ed. Windholz, M., Merck, Rahway, NJ, 9th edn., 1976, pp. 1287–1288. Paper 7/08256H Received November 17, 1997 Accepted February 27, 1998 Analyst, May 1998, Vol. 123 813

ISSN:0003-2654    

DOI:10.1039/a708256h

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Complementary use of capillary gas chromatography–mass spectrometry (ion trap) and gas chromatography–inductively coupled plasma mass spectrometry for the speciation of volatile antimony, tin and bismuth compounds in landfill and fermentation gases† Jörg Feldmann*a, Iris Kochb and William R. Cullenb a Department of Chemistry, University of Aberdeen, Meston Walk, Old Aberdeen, UK AB24 3UE b Department of Chemistry, University of British Columbia, 2036 Main Mall, Vancouver BC, Canada V6T 1Z1 ICP-MS is very sensitive and has limited matrix effects when used as an element-specific detector for GC in order to identify volatile metal or metalloid species.GC–MS is not very sensitive or selective in the electron ionization (EI) mode, but provides molecular information about volatile species. In this work, an ion trap EI-MS–MS and an ICP-MS system were used as two different detectors for the same GC system to provide complementary information about volatile organometallic species in the complex matrices of landfill and sewage sludge fermentation gases.A simple robust GC separation method with cryotrapping was adequate for the separation of the different metal(loid) containing volatile compounds, and was directly coupled to the ICP-MS system. In addition, gas samples from this GC system were collected in evacuated vials. These fractions were further separated on a capillary column and detected in an ion trap mass spectrometer. For the first time, parent ions, fragmentation patterns, isotopic ratios for Sb and Sn, and MS–MS data were used to identify positively Me3Sb, Me4Sn and Et2Me2Sn in landfill gas and Me3Sb and Me3Bi in fermentation gas.Keywords: Landfill gas; fermentation gas; gas chromatography–inductively coupled mass spectrometry; ion trap mass spectrometry; volatile organometallics; trimethylbismuthine; trimethylstibine; volatile organotin compounds In general, volatile metal and metalloid compounds can be generated in anthropogenic processes such as charging batteries (e.g., stibine),1 in the marine environment (e.g., stannanes2 and selenides3), in the soil by microorganisms (e.g., methylated arsines4 and stibines5) or as dimethyl selenide in human breath.6 Recently, we have shown, by using a purge and trap system coupled with ICP-MS, that the presence of a multitude of different volatile organic compounds in landfill and fermentation gas does not affect the detection of volatile metal(loid) species.7 Thus, methylated metal(loid) species such as MexAsH32x (x = 1–3), Me2Se, Me3Sb, Me4Sn, Me2Te, MeI, Me2Hg, EtxMe42xPb (x = 0–4), Me3Bi and the hexacarbonyls of Mo and W were separated by using this simple GC–ICP-MS system.8,9 Although molecular information about the volatile species is lost by using ICP-MS, the element-specific detection capabilities of this instrument allow a reasonably certain identification of the metal species by matching the retention times of unknown peaks with those for standards.However, species such as Me3Sb and Me3Bi in particular, which have been found for the first time in the environment, require verification of the molecular structure. The aim of this paper is to show the effectiveness of the complementary use of ICP-MS and EI-MS–MS as different detection methods for the GC separation of volatile metal and metalloid species in gases with a complex matrix. Experimental Chemicals and standards Me3SbCl2 was synthesized as described elsewhere.10 Me4Sn (98%, Alfa, Johnson Matthey, Ward Hill, MA, USA) was dissolved in methanol at 1000 mg l21 as Sn and diluted daily to 1 mg l21.Methyltin trichloride (Alfa Product), dimethyltin dichloride (Peninsular Chemresearch, Gainesville, FL, USA), trimethyltin hydroxide (Alfa Inorganic, Beverly, MA, USA), butyltin trichloride (Aldrich, Milwaukee, WI, USA) and dibutyltin dichloride (M&T Chemicals, Rahway, NJ, USA) were diluted in hexane to 1000 mg l21 as Sn and in de-ionized water to 1000 mg l21 as Sn.NaBEt4 (98%, Strem Chemicals, Newburyport, MA, USA) was diluted daily in de-ionized water to give a 1% m/m solution. NaBH4 (analytical reagent grade, Aldrich) was dissolved in de-ionized water and purged for 30 min with hydrogen daily to give a concentration of 6% m/m. The volatile standards in hexane were directly injected on to the capillary column (0.2 ml of 1000 mg l21 solution). Hydride generation and ethylation methodology were used to produce volatile standards from the partly alkylated tin and antimony compounds.Me3Sb was formed in the reaction of Me3SbCl2 with NaBH4. The reaction was performed at pH 7 in a 15 ml vial (closed off with a PTFE-faced silicone-rubber septum, 16 mm, Supelco, Bellefonte, PA, USA) by using 0.1 ml of Me3SbCl2 solution (30 ng as Sb) and 1 ml of de-ionized water. A 0.5 ml volume of 6% NaBH4 solution was injected through the septum.The headspace was sampled by using gas-tight syringes (1.0 ml, Gas tight No. 1001, Hamilton, Baton Rouge, LA, USA). MeSnH3, Me2SnH2, Me3SnH, BuSnH3, and Bu2SnH2 (200 ng as Sn) were generated under neutral conditions following the same procedure as for Me3Sb. Me2Et2Sn, MeEt3Sn, and Me3EtSn were generated by using the ethylation procedure according to Moens et al.11 A 100 ml volume of standard solution (1000 mg Sn ml21) in hexane was added to 5 ml of acetate-buffered water and 0.1 ml of NaBEt4 solution (1% m/m) in water was added.After shaking for 10 † Presented at The Third International Symposium on Speciation of Elements in Toxicology and in Environmental and Biological Sciences, Port Douglas, Australia, September 15–19, 1997. Analyst, May 1998, Vol. 123 (815–820) 815F1 F2 F3 4a 4b 3a 4b fitting syringe needle transfer line 100-120 °C –196 °C column He flow : heated with nichrome wire min, 1 ml of the hexane solution was injected on to the capillary column.Sampling procedure and sampling sites The gases from municipal waste deposits (Delta, Greater Vancouver Regional District, BC, Canada) and the gases from a mesophilic sewage sludge digester in a sewage purification plant (Iona Island, Vancouver, BC, Canada) are collected into gas wells and pumped in pipelines either to a furnace or to a power station. The gases in the gas wells were sampled directly into Tedlar bags by using a membrane pump (AirPro 6000D, Bios International, Pompton Plains, NJ, USA).Preconcentration and fractionation procedure The gas samples were cryogenically preconcentrated by trapping the gases on a U-shaped trap (31 cm 3 6 mm od), which was packed with Chromosorb (10% SP-2100, 45–60 mesh, 31 cm 3 6 mm od, Supelco) at 278 °C (dry-ice–acetone slush) (see Fig. 1). This relatively high temperature was chosen to avoid condensation of carbon dioxide and methane, the major components of landfill gas.12,13 No clean-up procedure or derivatization was applied to the gas samples in order to minimize changes in the molecular structure of volatile compounds.The preconcentration procedure included a combination of thermal desorption of the cryotrapped sample and separation by using a non-polar chromatographic column. The column was heated with a Nichrome wire (8 V, 10 A) from 278 to 150 °C over a period of 3 min and the gases were separated by using a helium flow rate of 133 ml min21.The separated sample was transported through a heated Teflon transfer line (1.1 m 3 3 mm od) to a syringe (1 ml, polypropylene). This syringe was cut at the top and attached to the transfer line with a Swagelok reducing unit (PTFE, from 3 to 6 mm). A stainlesssteel needle was attached to the end of the syringe. Evacuated glass vials (15 ml), which were closed off with silicone-rubber PTFE septa, were employed to collect the gas fractions. The fractions were taken every 7 s to fill the evacuated 15 ml vials completely.The sampling frequency was nine fractions per minute . The sampling time for the first fraction (F1) started at 120 s. Twenty-five fractions were sampled; consequently, the last fraction (F25) was finished at 295 s. GC–MS–MS A GC–MS–MS system consisting of a Star 3400Cx gas chromatograph, equipped with a Model 1078 temperature programmable injector and interfaced to a Saturn 4D ion-trap mass spectrometer (Varian, Palo Alfo, CA, USA) was employed for this analysis.Gas-tight syringes (Gastight No. 1001, Hamilton) were used to inject the gas and the headspace samples on to a capillary column [PTE-5, 30 m 3 0.32 mm id, 0.25 mm, Supelco 2-4143, poly(5% diphenyl–95% dimethylsiloxane)] after they had been rinsed for 20 s with laboratory air. The injector was kept at 100 °C. The temperature programme was from 40 to 150 °C at 15 °C min21. The mass range was chosen according to the masses expected for the compounds being analysed.The electron ionization (EI) mode was used. The parameters used (Tables 1–3) were optimized to achieve the highest signal-to-noise ratio of the major fragment by using the trimethylstibine standard (headspace hydride generation). The MS–MS option was used to generate a second fragmentation pattern to confirm further the identity of specific masses. GC–ICP–MS The procedure was given in detail in a previous paper.9 Briefly, the procedure described above for the fractionation was followed but the syringe with the needle was replaced by a PFA fitting that physically coupled this GC system on-line to the ICP-MS system (Fig. 1). In addition to the GC flow, an aqueous solution of Rh (10 ppb in 1% HNO3) was introduced as a wet aerosol into the plasma by using a De Galan nebulizer. Both gas flows were mixed together in a tee-piece (6 mm od) inserted between the spray chamber and the torch, replacing the quartz elbow usually in this position. The same GC parameters were used preceding ICP-MS detection as were used for the fractionation of the gas samples.The operational parameters of the VG Plasmaquad PQ2 Turbo (VG Elemental, Winsford, Cheshire, UK) were similar to those used when ICP-MS is used for routine water analysis. A cooling gas flow rate of 13 l min21, Fig. 1 Schematic diagram of the GC fraction collector, showing alternative configuration with ICP-MS detection (dotted lines). Table 1 Ion trap MS parameters Mass range m/z 100–200 or 200–330 Scan time 0.4 s Segment length 8 min Peak threshold 0 counts Mass defect 0 mm / 10 Background m/z 99 or 150 Ion mode Electron ionization Ion preparation MS–MS or none Ion control Auto Multiplier 2150 V Target 40 800 or 10 000 (MS–MS) Ionization current 20 mA Manifold temperature 260 °C MS–MS mode Collision ionization detection Waveform type Non-resonance Amplitude 20, 30, 50, 70 V Mass isolation window m/z 1.0 Table 2 Capillary GC (CGC) parameters used Injector temperature 200 °C Column temperature programme From 40 to 150 °C at 15 °C min21 Transfer line temperature 200 °C Column PTE-5, 30 m 3 0.32 mm id, 0.25 mm, Supelco 2-4143 Table 3 GC parameters used for preconcentration and for on-line coupling to ICP-MS Column temperature 2196 to +150 °C Cryogenic supply Liquid nitrogen Carrier gas Helium Joint unit Tee-piece, quartz-pipe, 6 mm od Flow rate 133 ml min21 Column 22 cm 3 6 mm od, 10% SP-2100 on Chromosorb Transfer line PTFE tube, 100 cm 3 3 mm od, 120 °C 816 Analyst, May 1998, Vol. 123an auxiliary gas flow rate of 0.7 l min21 and a nebulizer gas flow rate of 1.05 l min21 were chosen. The added 10% He from the gas chromatography did not significantly influence the plasma. The ICP-MS parameters were optimized to give similar sensitivities for Sn, Sb, and Bi by using a 0.1% HNO3 tune solution of 10 ng ml21 Sn, Sb and Bi. A Henry generator was used with a forward power of 1350 W, resulting in a reflected power of < 7 W.Results and discussion GC–ICP-MS Fig. 2 shows a GC–ICP-MS trace for a landfill gas sample, which contains at least five different volatile tin compounds (A– E), one antimony compound (F) and one bismuth compound (G). The presence of these species was revealed in our earlier studies that used GC–ICP-MS.7,8,14 The compounds were identified by using element-specific detection and by matching the retention times with those of standards. Table 4 gives the retention times and the calculated boilingpoints of these unknowns compared with the boiling-points of volatile tin, antimony and bismuth standards.Isotopic fingerprints can also be used for the identification of elemental tin and antimony, because both have more than one isotope. To do so, each chromatogram (for each m/z in the range 110–126) was integrated for the landfill sample and standards. The sum of all isotopes for each element was normalized, so that the amount of each m/z could be expressed as the relative abundance of the isotope (Fig. 3). The isotopic fingerprints of tin and antimony from a landfill sample are compared to those for standards in Figure 3. The similarity between the tin and antimony isotopic fingerprints in the landfill sample and the standards shows unequivocal evidence of the presence of volatile compounds containing these elements. No isotopic fragmentation is known for these heavy elements. Identification of monoisotopic elements such as bismuth is not as clear, but can be improved by detecting possible interferences (e.g., 208PbH+ on m/z 209 and 207PbH+, 208Pb on m/z 208).At the retention time of 160 s, where m/z 209 (elemental Bi) was detected, no other peak occurred that would result in detection at this mass. However, analysis with a second method is necessary to confirm that the peak is due to Bi and not an interfering ion. GC–MS To confirm further the identification of the volatile species, a non-destructive ionization method and better chromatographic resolution were needed.The analytical column used was not sufficient to distinguish between mixed alkylated tin com- Fig. 2 Chromatograms of tin, antimony and bismuth that were determined on m/z 120, 121, and 209 in landfill gas simultaneously by using GC–ICPMS. A, Me4Sn; F, Me3Sb; G, Me3Bi; B–G, higher alkylated (butylated or ethylated) tin compounds. Table 4 Species and retention time of the unknown peaks and their calculated boiling-points compared with those of volatile tin, antimony and bismuth standards with the same retention time as the unknown peaks Fraction Retention number and Species time/s Calc.bp °C* Std. bp °C sampling time/s SnA 137 81 78 (Me4Sn) F3 (134–141) SnB 165 112 108 (EtMe3Sn) F7 (162-169) 100 (BuSnH3) SnC 198 141 145 (Et2Me2Sn) F12 (197-204) SnD 234 166 163 (Et3MeSn) F17 (232-239) SnE 285 190 181 (Et4Sn) F24 (281-288) 200† (Bu2SnH2) SbF 135 78 81 (Me3Sb) F3 (134-141) BiG 160 106 109 (Me3Bi)‡ F6 (155-162) * Bp = 0.0005 3 1026 t3 2 0.0059 t2 + 2.5309 3 2167.58; r2 = 0.9972.† decomposed earlier, therefore estimated bp. ‡ The same retention time as the unknown peak in sewage and landfill gas given by Feldmann et al.14. Fig. 3 Isotopic fingerprint of tin and antimony detected simultaneously in landfill gas compared with the relative abundance of naturally occurring isotopes, measured in Me4Sn and Me3Sb standards. The sum of all isotopes for each element is normalized to 100%.Analyst, May 1998, Vol. 123 817pounds and alkylated tin hydrides (e.g., EtMe3Sn and BuSnH3). The resolution for this packed column, calculated for Mo(CO)6 and W(CO)6 by using their retention times and their peak halfwidths, was 3.84, contrasting with a resolution for the 30 m capillary column of 19.26. However, the detection limits of the volatile metal(loid) species are much better for the packed column GC–ICP-MS method than for the capillary column GC– MS method.Using GC–ICP-MS, the LOD for the different volatile metal(loid) species were in the pg ml21 range, without any optimized separation technique and without any clean-up techniques, because the ICP is able to destroy most of the interferences from other organic compounds in a preconcentrated gas sample, as shown earlier with a similar system.15 Me3Sb and MexSnH42x (x = 1–3) were generated by using hydride generation methodology. When 1 ml of the headspace was injected into the GC–MS system a detection limit of 0.2 ng of Sb for Me3Sb was obtained, corresponding to 1 ng of Sb in solution before derivatization. However, the analysis of the headspace following hydride generation suffers from imprecision, since RSD values no better than 20% could be obtained for five replicate analyses.Fig. 4 shows the mass spectra of the standard trimethylstibine (3.8 ng Sb as Me3Sb); however, in this case, the standard was not hidden in a matrix of many volatile organic compounds.The direct injection of 1 ml of landfill or fermentation gas without any preconcentration and fractionation was not successful in identifying volatile metal(loid) species. When using the GC fractionation procedure, the previously evacuated 15 ml vials were filled in 7–8 s. A concentration gain of approximately 1000-fold was achieved when 15 l of gas were cryotrapped and fractions were taken for 7 s, corresponding to analyte peak widths of approximately 7–8 s.Therefore, the organometallic species in landfill gas which are usually determined to be in the pg ml21 concentration range (Delta: Me3Sb 4.08–17.1, Me4Sn 24.0–28.4, Me3Bi 0.013–0.030 pg ml21) in mainly methane and carbon dioxide, were preconcentrated in the vial in the ng ml21 range in helium. The same chromatographic conditions were used for the fractionation as for the GC–ICP-MS method, allowing sample collection of fractions that contain detected tin, antimony and bismuth species (Table 4).Me4Sn and Me3Sb were identified in landfill gas by matching the retention times and isotopic fingerprints of the peaks in samples and standards when using GC–ICP-MS. The retention times (47 s for Me4Sn and 48 s for Me3Sb) achieved using CGC–MS also match the times for standards in fraction F3, which is the fraction that should contain these compounds according to the GC–ICP-MS traces. The mass spectra of fraction F3 at 47 s [Fig. 6(b)] and 48 s [Fig. 5(c)] show the same fragmentation pattern as the mass spectra for the standards Me3Sb (Fig. 4) and Me4Sn [Fig. 6(a)]. The isotopic fingerprint obtained from the EI mass spectra can be used to identify Sb and Sn in these compounds also (Table 5). However, the ratios of m/z 121 and 123 for Sb and m/z 118 and 120 for Sn show slightly different values than the naturally occurring isotopes, because of the presence of interfering fragments not found when using Fig. 5 (a) GC–ICP-MS trace for 100 ml of landfill gas monitored for antimony at m/z 121. After fractionation the preconcentrated fraction F3 was injected into the CGC–MS system. (b) TIC spectrum (m/z 115–m/z 175). The X indicates the retention time of fragmentation pattern shown in (c). Fig. 4 Mass spectrum of trimethylstibine (48 s) by using CGC–MS. Me3Sb were generated after using the hydride generation method in a vial, where 50 ng of Me3SbCl2 was added and 7.7% of the headspace (1 ml) was injected. 818 Analyst, May 1998, Vol. 123ICP-MS detection. In the positive ion mode, the favoured mode for these compounds where the so-called ‘organometallic’-type fragmentation takes place,17 the mass spectra of Me3Sb and Me4Sn show substantial loss of a CH3 group. The major fragments are always M2CH3 +, and for Me4Sn no M+ peak was observed, probably because no positive charge can be located on Me4Sn, in contrast to Me3Sb. The spectra show also the loss of H of the CH3 group; thus fragments were found that were one mass unit lower than (M2CH3)n +.Although no Sb–H bond was present in the original species, this fragment, at m/z 122 and 124, was found in both the standard and the samples. Rearrangement and migration of H atoms are characteristic of these molecules. In fraction F12, which contained the unknown tin species Snc, an unknown peak at a retention time of 81 s was detected by using CGC–MS in the EI mode [Fig. 7(b)]. The fractionation pattern for this peak is similar to that for the Et2Me2Sn standard [Fig. 7(a)], which had the same retention time after being generated by using the ethylation procedure. The small differences can be caused by the matrix gases in the sample, which were also injected into the ion trap and may alter the ionization. No other volatile tin compounds or the volatile bismuth compound could be identified in the landfill gas by using capillary GC–MS. The concentration of total volatile species found in fermentation gas from Iona Island were similar to those determined in gases from a German sewage water purification plant.8 Me4Sn could be identified by using GC–ICP-MS but not with the ion trap, because of the three orders of magnitude lower concentration in this sample, compared with the previous study.No other tin species were detected with GC–ICP-MS or with CGC–MS. The amount of Me3Sb was the same order of magnitude as the amounts found in gases from the German plant, but it was not quantified precisely.The GC–ICP-MS trace and the fragmentation pattern from CGC–MS for Me3Sb after fractionation were similar to those of the landfill gas sample shown in Fig. 5(c). One volatile bismuth species was found at concentrations of 0.01–0.03 ng l21 in the landfill gas and in an amount at least three orders of magnitude higher in the fermentation gas by using GC–ICP-MS. In a previous study,14 a synthesized Me3Bi was analysed immediately after distillation and its retention time was the same as that of the unknown Bi peak in the landfill gas by using GC–ICP-MS.By using the preconcentration procedure almost all the Bi present was sampled in fraction F6 (155–162 s). No identification of a bismuth species was possible in any fraction of the landfill gas by using CGC–MS–MS. However, the fermentation gas shows the occurrence of a bismuth species in this fraction. Fraction F6 of the fermentation gas contains higher amounts of the volatile Bi, so that an unambiguous mass spectrum was produced, which shows a fragmentation pattern similar to that of Me3Sb [Fig.8(a)].Since Bi is monoisotopic, identification by MS in the presence of interferences is difficult. In addition to the parent ion (m/z 254), the loss of CH3 results in m/z 239 and m/z 224. However no Bi+ Fig. 6 (a) Mass spectrum of 117 ng of tetramethyltin. (b) Fragmentation pattern of the landfill gas fraction F3 at 47 s. Table 5 Isotopic fingerprints of the detected volatile species in landfill gas using ion trap MS data Sb CH3Sb (CH3)2Sb (CH3)2Sb-CH2 (m/z 121/123) (m/z 136/138) (m/z 151/153) (m/z 165/167) Sample 1.52 1.53 1.70 1.64 (CH3)3Sb standard 1.55 1.55 1.44 1.55 Ref. 16 1.34 Sn MeSn Me2Sn Me3Sn (m/z 118/120) (m/z 133/135) (m/z 148/150) (m/z 163/165) Sample 0.74 0.67 1.44 0.80 Me4Sn standard 0.89 0.95 1.11 0.88 Ref. 16 0.74 Fig. 7 (a) Mass spectrum of diethyldimethyltin, generated by using the ethylation procedure and liquid–liquid extration into hexane.(b) Fragmentation pattern of the landfill gas fraction F12 at 81 s. Analyst, May 1998, Vol. 123 819(m/z 209) was found as a fragment; instead, m/z 210 was observed. According to the fragmentation pattern of Me3Bi at 30 eV, Bi+ at m/z 209 was formed as the major fragment in addition to the other species at m/z 224 (70%), m/z 239 (85%) and m/z 254 (50%).18 To investigate if the fragment at m/z 210 contains Bi, the MS–MS mode was used. Only mass 210 was stored in the ion trap as a parent ion and fragmented by applying different voltages (by changing the amplitudes from 20 to 70 V, see Table 1).The major fragmentation peak was at m/z 209 in a significant amount (the impurities in the spectrum are probably from lead compounds). Therefore, when using this method, one of the major fragments of trimethylbismuthine is BiH+, which can be expected, since a similar H· migration was found for trimethylstibine to give SbH+.Conclusions The complementary information from these two systems, summarized in Table 6, allows the characterization of volatile organometallic compounds in environmental samples such as landfill gas and fermentation gas. Cryotrapping and coarse separation by using a non-polar packed column guarantees that the species of interest are not modified when the ICP-MS system is used as a very sensitive element-specific detector. This same coarse separation as a preconcentration and fractionation step allows the unequivocal identification and characterization of Me3Sb, Me4Sn and, for the first time, Me3Bi in fractions of landfill and fermentation gas by using their retention times on a capillary column, the isotopic fingerprints and the MS and MS–MS data. We gratefully acknowledge financial support from the Alexander von Humboldt Foundation (AvH) and NSERC Canada.We also thank Mr. A. Mosi and Dr. C. Simpson for their help.References 1 Hetland, S., Martinsen, I., Radzuik, B., and Thomassen, Y., Anal. Sci., 1991, 7, 1029. 2 Donard, O. F. X., and Weber, J. H., Nature (London), 1988, 332, 339. 3 Amouroux, D., and Donard, O. F. X., Geophys. Res. Lett., 1996, 23, 1777. 4 Cheng, C. N., and Focht, D. D., Appl. Environ. Microbiol., 1979, 38, 494. 5 Gürleyük, H., van Fleet-Stadler, V., and Chasteen, T. G., Appl. Organomet. Chem.,1997, 11, 471. 6 Feldmann J., Riechmann, T., and Hirner, A. V., Fresenius’ J.Anal. Chem., 1996, 354, 620. 7 Feldmann, J., Grümping, R., and Hirner, A. H., Fresenius’ J. Anal. Chem., 1994, 350, 228. 8 Feldmann, J., and Hirner, A. V., Int. J. Environ. Anal. Chem., 1995, 60, 339. 9 Feldmann, J., and Cullen, W. R., Environ. Sci. Technol., 1997, 31, 2125. 10 Morgan, G. T., and Davies, G. T., Proc. R. Soc. London, Ser. A., 1926, 110, 523. 11 Moens, L., De Smaele, T., Dams, R., Van Den Broeck, P., and Sandra, P., Anal. Chem., 1997, 69, 1604. 12 Ward, R. S., Williams, G. M., and Hills, C. C. Waste Manage. Res., 1996, 14, 243. 13 Hirner, A. V., Feldmann, J., Goguel, R., Rapsomanikis, S., Fischer, R., and Andreae, M. O., Appl. Organomet. Chem., 1994, 8, 65. 14 Feldmann, J., Krupp, E. M., Müller, L. M., and Hirner, A. V., Appl. Organomet. Chem., to be submitted. 15 Feldmann, J., J. Anal. At. Spectrom., 1997, 12, 1069. 16 De Bièvre, P., and Barnes, I. L., Int. J. Mass Spectrom. Ion Processes., 1985, 65, 211. 17 Nekrasov, Y. S., and Zagorevskii, D. V., in The Chemistry of Organic Arsenic, Antimony and Bismuth Compounds, ed. Patai, S., Wiley, Chicester, 1994, pp. 237–264. 18 Kostyanovskii, R. G., and Plekhanov, V. G., Org. Mass Spectrom., 1972, 6, 1183. Paper 7/07478F Received October 16, 1997 Accepted January 12, 1998 Table 6 Summary of the advantages and limitations of GC–MS (ion trap) and GC–ICP-MS method for the determination of volatile metal(loid) compounds in landfill and fermentation gas Parameter CGC–MS GC–ICP-MS Sample volume needed ~ 100 l 0.1–1 l Detection limits (3s) 0.2 ng Me3Sb 0.3 pg Me3Sb Specificity Interference from volatile organic and organometallic compounds Limited interference; Elementspecific; in addition isotopic fingerprint Resolution 19.26 3.84 Structural information Fragmentation pattern in addition MS–MS option Retention time comparison with standards, no structure information Time Off-line fractionation; and online separation analysis time: 1–2 h On-line separation, analysis time 6 min Fig. 8 (a) Fragmentation pattern at 60 s of the fractionated fermentation gas fraction F6. (b) MS–MS trace obtained using 70 V, after storing m/z 210 in the trap. 820 Analyst, May 1998, Vol. 123

ISSN:0003-2654    

DOI:10.1039/a707478f

出版商:RSC    

年代:1998

数据来源: RSC