Determination of Trace and Ultratrace Elements in Human Serum With a Double Focusing Magnetic Sector Inductively Coupled Plasma Mass Spectrometer† JO� RGEN RIONDATO, FRANK VANHAECKE, LUC MOENS AND RICHARD DAMS* L aboratory of Analytical Chemistry, Ghent University, Institute for Nuclear Sciences, Proeftuinstraat 86, B-9000 Ghent, Belgium The purpose of this study was to evaluate the potential of gases introduced into the plasma, water and the reagents used. In addition, biological samples with a heavy matrix can ICP-MS with a double focusing instrument for the determination of both spectrally and not spectrally interfered contribute to an important extent to the formation of polyatomic ions. Consequently, a significant number of elements ultratrace elements in human serum after minimised sample pre-treatment and without using any separation or cannot be determined without preliminary work, aimed at a reduction or elimination of spectral interferences.The latter preconcentration techniques.Because of the heavy matrix (6–8% proteins and 1% inorganic compounds) of the material can, for example, be obtained by reducing the amount of solvent reaching the plasma1 or by applying electrothermal investigated, a higher resolving power than is available with quadrupole instruments was necessary for the determination of vaporisation for sample introduction.2–5 The most elegant and direct solution, however, is to use a double focusing magnetic some elements in order to eliminate spectral interferences.This was achieved by using ICP-MS apparatus equipped with a sector field mass spectrometer instead of a quadrupole filter. The ICP-MS instrument used in this study, the Element double focusing magnetic sector MS. This type of instrument also permits lower detection limits when compared with from Finnigan MAT (Bremen, Germany) allows operation in three dierent resolution modes (m/Dm=300, 3000 and 7500, traditional quadrupole ICP-MS instruments, mainly owing to the low instrumental background values.Determinations were 10% valley definition). In this study, the potential of the apparatus has been proved by analysing the Second Generation carried out using external calibration and, for most of the elements, also single standard additions as calibration Human Serum Reference Material.6 At the low resolution setting, a combination of high sensitivity and an extremely low techniques.Both the non-spectrally interfered (Cd, Sn, Ag and U) and the spectrally interfered (Al, Si, P, S, Ti, Cr, Mn, Fe, background permits detection limits significantly lower than those obtained by quadrupole-based ICP-MS. In order to Cu and Zn) elements were determined in a Second Generation Human Serum Reference Material. Sample preparation determine those elements for which the nuclides of interest are subject to spectral overlap at low resolution, a higher resolution consisted in reconstitution of the freeze-dried material or in microwave digestion, both followed by dilution.In addition to setting is necessary. Although this implies a decrease in sensitivity, the resulting detection limits are generally suciently continuous nebulization, flow injection was also applied as an introduction technique for the determination of some elements low.7 Previous analysis of the reference material by quadrupole in reconstituted serum. The results for the certified elements were all within the certified range except for Cr for which a ICP-MS8 enabled the determination of the concentrations of up to 15 trace and ultratrace elements.Because of the aforemen- significantly higher value was obtained. The results of the remaining elements were compared with literature values tioned limitations of this technique, some of the elements of interest could not be determined. For the serum analysis where they existed. Very low concentrations could be determined: for instance a concentration as low as 98 ng l-1 described in the present work, chemical separation or other preliminary sample preparation was not required except for a was determined for Ag with a standard deviation of less than 10%.The major diculty encountered in this study was to four- or five-fold dilution with 0.14 M HNO3, which seemed to be sucient to minimise signal suppression and other matrix keep the blank values suciently low, which could only be achieved by using ultra-pure reagents and by working in a eects.9 Some elements, however, have also been determined in microwave digested samples and results were compared.clean environment. Finally, a flow injection system was used to introduce the Keywords: Ultratrace elements; spectral interference; double (diluted) reconstituted serum samples for some of the elements focusing inductively coupled plasma mass spectrometry; human of interest. serum; flow injection; blank value EXPERIMENTAL ICP-MS is a powerful technique for trace and ultratrace High Resolution ICP-MS Instrumentation analysis of biological samples and especially of body fluids, combining rapid analysis with excellent detection limits and Measurements were carried out on an Element ICP mass multi-element capabilities.However, most instruments are spectrometer from Finnigan MAT. Details about the instruequipped with a quadrupole mass filter which implies that mentation used and its capabilities are to be found elsespectral interferences often jeopardise accurate trace element where.7,9–11 The instrumental operating conditions are determination, especially at mass-to-charge ratios of 80.Most summarised in Table 1. of these interferences are caused by polyatomic ions originating from a combination of elements from the plasma, atmospheric Flow Injection Instrumentation The flow injection system used is a modular system purchased † Presented at the 1997 European Winter Conference on Plasma Spectrochemistry, Gent, Belgium, January 12–17, 1997.from EVA (EVA product line is a joint venture of FIAtron Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 (933–937) 933Table 1 Instrumental operating conditions 10 mg l-1, except for P and S, at the 100 mg l-1 level ). However, 115In was used as an internal standard at a 1 mg l-1 level Rf power 1370 W for the determination of Ag, Cd and Sn and finally 205Tl Gas flow rates (1 mg l-1) acted as an internal standard for U.Outer plasma 14 l min-1 Standard solutions were prepared containing 1–1000 mg l-1 Intermediate 0.7 l min-1 Aerosol carrier 0.7–0.9 l min-1 depending on the isotopic abundance and the ionisation energy Sample uptake rate 1 ml min-1 of the nuclide of interest. The concentration level chosen also Ion sampling depth Adjusted in order to obtain maximum varied with the mass-to-charge ratio (m/z) depending on signal intensity the sensitivity of the instrument.7 In addition, the standard Ion lens settings Adjusted in order to obtain maximum solutions of spectrally interfered elements contain at least signal intensity and resolution 10 mg l-1 since a resolution of 3000 implies a decrease in sensi- Sampling cone Nickel, 1.1 mm orifice diameter Skimmer Nickel, 0.8 mm orifice diameter tivity by approximately a factor of 20 compared with low Torch Fassel type resolution (300).7 Nebulizer Meinhard TR-30-A3 Spray chamber Scott-type (‘double pass’), cooled to 5 °C Calibration Techniques All determinations were carried out using external calibration Systems, Oconomowoc, WI, USA and Eppendorf-Netheleras a calibration technique.For some of the elements which Hinz, Hamburg, Germany). The sample loop with a volume were determined in the five-fold diluted samples the method of 200 ml was controlled by a switchable valve allowing two of standard additions was applied as well.positions (loading the loop or injecting the sample). The carrier solution (0.14 M HNO3) was pumped at 1.2 ml min-1 while the introduced sample solution was pumped at 0.7 ml min-1. Samples and Sample Preparation The Second Generation Human Serum Reference Material analysed contains 14 trace elements with certified concen- Acquisition trations.6 This reference material has been prepared under The elements for which determination was not hampereas a rigorously controlled conditions to avoid any contamination result of spectral interferences were determined using a resoor losses of trace elements and the trace element concentrations lution setting of 300.The other elements required a resolution therefore closely approximate those in normal human serum. setting of 3000 to separate the signals of the analyte and the Some of the elements determined in this work were present at interfering polyatomic ions. All measurements were performed very low concentrations and were not certified; for U no in the electric scan (E-scan) acquisition mode.literature values have been reported until now. As previously As indicated in Table 2, for at least three independently mentioned, the freeze-dried serum was prepared for analysis prepared sample solutions, three replicate measurements each in two dierent ways. Simple reconstitution was obtained by of 1 min per isotope were carried out. However, for determiadding 6 ml of Millipore Milli-Q water to 0.5 g of the lyonation with flow injection, five replicate acquisitions of only philised serum which was transferred quantitatively into a 30 s each were carried out, owing to the transient nature of calibrated flask (25 ml for five-fold dilution) and adjusted to the signals obtained (signal duration typically 10–15 s).Data volume with 0.14 M HNO3. This sample preparation method were acquired during a short period of time half way through has the advantages of being less time-consuming and introducthe transient signal, where this is more or less constant.Blanks, ing less contamination. On the other hand, the heavy matrix samples and standards were measured in that order. was seen to cause a rapid and irreversible clogging of the torch. Therefore, for some of the elements, the reference material was also prepared for analysis by microwave digestion Reagents and Standard Solutions and the results obtained were compared. For this digestion, a In order to reduce the risk of contamination, all work was commercially available microwave oven (Amana Radarange, carried out under clean room conditions.model RS 560A, Amana Refrigeration International, IA, USA) All standard solutions used were prepared by successively was used. First, 5 ml of HNO3 (purified by sub-boiling distildiluting commercially available mono-element standard solulation) were added to #0.5 g of serum in a PFA digestion tions (1 g l-1). For these dilutions, 100-fold diluted (0.14 M) bomb.The destruction was achieved by subjecting the bomb HNO3 was used. The HNO3 (14 M) was purified by sub-boiling to a four-step heating programme, during which an increasingly distillation and diluted with Millipore Milli-Q water. In order higher power was applied. Again, the solution obtained was to avoid contamination, only thoroughly precleaned polyethyltransferred quantitatively into a 25 ml calibrated flask and ene vials were used and dilutions were carried out using adjusted to volume with 0.14 M HNO3.Attention should be micropipettes. paid to avoid contamination from vessels and especially from For the determination of Al, Si, P, S, Ti, Cr, Mn, Fe, Cu the HNO3 used. and Zn, 59Co was used as an internal standard (at a level of In Table 3 blank serum values for several elements are compared with their concentrations in the measured diluted Table 2 Measurement parameters Table 3 Average serum blank concentrations and concentrations in Acquisition mode Electric scanning the diluted Human Serum Reference Material (blank corrected) Number of samples 3 Number of replicate measurements per sample 3* Element Serum blank Diluted serum 5† Number of scans per replicate 60* Al* 0.73 mg l-1 0.34 mg l-1 Al† 0.85 mg l-1 0.30 mg l-1 300† Time per scan 1 s* Si* 15 mg l-1 35 mg l-1 Ti* 0.05 mg l-1 0.22 mg l-1 0.1 s† Resolution setting m/Dm 300 and 3000 Ag† 5 ng l-1 18 ng l-1 * Continuous nebulization.* Sample preparation: reconstitution only. † Sample preparation: microwave digestion. † Flow injection. 934 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12human serum (Reference Material). As can be observed, blank Ag values for some elements largely exceed the instrumental Determination of the element Ag (not certified in the reference background making it impossible to take full advantage of the material ) has also been carried out and published in a previous extremely low instrumental detection limits of the Element publication.12 The 107Ag nuclide (51.8%) was used for the system.determination. Since no spectral interferences occur at a massto- charge ratio of 107, a resolving power of 300 was sucient. The resulting value is substantially lower than the only value RESULTS AND DISCUSSION reported for this material and obtained by RNAA13 (see Determination of Trace Elements in Diluted Reconstituted Table 6).The relatively high standard deviation is caused by Human Serum a high blank value of 38.5 ng l-1. The Ag concentration measured in the four-fold diluted serum corresponds to about As mentioned before, some of the elements of interest (Ti, Fe, 15 ng l-1, which is below the detection limit attainable with a Cu, Zn, Ag, Cd, Sn and U) were determined using both quadrupole ICP-MS. external calibration and single standard additions as calibration techniques. Excellent agreement between the results (within 5%) from both techniques was observed, implying that Mn and Cr internal standardisation is sucient to overcome matrix eects.The concentrations of Mn and Cr in the reference material are Indeed, for all the elements measured only a slight or even no low (<1 mg l-1) and in addition, all nuclides involved are signal suppression was observed in the diluted serum. spectrally interfered, making determination with quadrupole ICP-MS impossible. In this work, Cr was determined at a Fe, Cu and Zn resolution of 3000 via the main 52Cr nuclide; the peak of 52Cr+ had to be measured next to a 1000 times more intense ArC+ Although in the measurement of diluted human serum all the peak.Manganese is mono-isotopic (55Mn) but polyatomic ions isotopes of these elements are interfered to a significant extent, with a mass-to-charge ratio of 55 (Table 4) can easily be determination via quadrupole ICP-MS has been carried out separated from the Mn+ ion signal at R=3000.The Mn result previously at this laboratory.8 Matrix matching of the blank was in excellent agreement with the certified value.6 The Cr after a thorough study of the composition of the serum matrix result, however, seemed to be high, probably due to and an appropriate choice of the (less interfered) nuclides to contamination. be measured made accurate quadrupole determinations possible. With a double focusing instrument, set at a resolution of 3000, the peaks of the isotopes 56Fe, 63Cu and 66Zn were P and S separated from the interferences listed in Table 4.These In order to determine these elements present at a concentration elements were determined using a prototype instrument with of up to 100 mg l-1, 50-fold diluted samples were measured. limited instrumental and software possibilities. The results In spite of the fact that P and S are highly concentrated and obtained have already been reported in a previous publiconsequently are not considered as trace elements, until now cation12 and are shown in Table 5.Good agreement with determination with ICP-MS was problematic or impossible certified values was obtained for Fe and Cu whereas the Zn because of large spectral interferences. A spectrum at a mass- result was in reasonable agreement. to-charge ratio of 32 is presented in Fig. 1 for the Human Serum Reference Material showing the 32S+ signal next to the Table 4 Nuclides used for element determination and the most 16O2+ peak.important polyatomic ions, jeopardising accurate determination at low resolution in human serum Determination of Trace Elements in Microwave-digested Element Nuclide Interfering species Human Serum Aluminium 27Al (100%) 13C14N, 12C14N1H Silicon 28Si (92.2%) 12C16O, 14N2 To overcome the rapid clogging of the torch when analysing Phosphorus 31P (100%) 14N16O1H, 15N16O simply diluted (human) serum, freeze-dried serum was also Sulfur 32S (95.0%) 16O2 microwave digested prior to the measurement.Titanium 47Ti (7.3%) 31P16O, 12C35Cl 49Ti (5.5%) 35Cl14N, 32S16O1H, 12C37Cl Chromium 52Cr (83.8%) 40Ar12C, 35Cl16O1H Fe, Cu, Zn and Ag Manganese 55Mn (100%) 39K16O, 40Ar14N1H Iron 56Fe (91.7%) 40Ar16O, 40Ca16O To allow comparison between both methods of sample prep- Copper 63Cu (69.1%) 40Ar23Na, 31P16O2 aration (digestion and simple dilution) Fe, Cu, Zn and Ag were Zinc 64Zn (48.9%) 32S16O2, 32S2 determined. The results for Fe and Zn in digested samples 66Zn (27.9%) 34S16O2, 34S32S were in agreement but the Cu result was 4% lower.Silver has Table 5 ICP-MS results (with standard deviations) for some certified elements in the Human Serum Reference Material in microwave digested and simply diluted (five-fold unless indicated otherwise) samples Nuclide used Resolution Digested Reconstituted Certified range 27Al 3000 1.49 (0.61) mg l-1 1.70 (0.54)* mg l-1 1.59–2.12 mg l-1 52Cr 3000 — 135 (25) ng l-1 61–79 ng l-1 55Mn 3000 — 701 (22) ng l-1 672.7–727.3 ng l-1 56Fe 3000 2.08 (0.14) mg l-1 2.153 (0.074)† mg l-1 2.21–2.49 mg l-1 63Cu 3000 0.932 (0.012) mg l-1 0.970 (0.021)† mg l-1 0.97–1.05 mg l-1 64Zn 3000 0.850 (0.053) mg l-1 — 0.85–0.89 mg l-1 66Zn 3000 — 0.750 (0.074)† mg l-1 0.85–0.89 mg l-1 111Cd 300 226.4 (4.5) ng l-1 — 154.5–227.3 ng l-1 * Determined with flow injection. † Eight-fold diluted; from ref. 12. Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 935Table 6 ICP-MS results (with standard deviation) for some non-certified elements in the Human Serum Reference Material in microwave digested and simply diluted (five-fold unless indicated otherwise) samples Nuclide used Resolution Digested Reconstituted (Lowest) literature values Reference 28Si 3000 — 175 (20)* mg l-1 170 mg l-1 14 31P 3000 — 122.7 (1.5)† mg l-1 115–163 mg l-1 15 32S 3000 — 1.113 (0.021)† g l-1 1.12–1.27 g l-1 15 47Ti 3000 — 0.99 (0.24)* mg l-1 1.3±0.3 mg l-1 16 0.7±0.1 mg l-1 16 49Ti 3000 — 1.09 (0.17)* mg l-1 1.3±0.3 mg l-1 16 0.7±0.1 mg l-1 16 107Ag 300 98.2 (2.7) ng l-1 53.6 (6.4)‡ ng l-1 84.5 (3.6) ng l-1 13 120Sn 300 601 (26) ng l-1 — 690.9 (127.3) ng l-1 17 238U 300 — 0.77 (0.14)* ng l-1 — * Determined with flow injection.† 50-Fold diluted. ‡ Four-fold diluted; from ref. 12. lution, indicating that the applied microwave destruction was not yielding complete mineralisation which was confirmed by earlier experiences in this laboratory.Analysis of Reconstituted Human Serum Using Flow Injection for Sample Introduction Finally, for the determination of Al, Si, Ti and U, flow injection was used as an introduction system. It allows small amounts of sample solution to be injected and the system to be rinsed with 0.14 M HNO3 in between injections. Thus deposition of matrix components in the torch, especially proteins, can be avoided. Also memory eects are reduced to a significant Fig. 1 Sulfur signal in 50-fold diluted human serum at a resolution extent.The only drawback of this technique is that the analyses setting of 3000. are somewhat more time-consuming. There is no significant dierence between the results of determinations via both sample been determined with a precision better than 10%; a signifi- introduction techniques. cantly higher concentration was however found than the value obtained in the reconstituted samples under non-optimal Al conditions.12 Results are shown in Tables 5 and 6.The result of the determination of Al in reconstituted samples (1.70 mg l-1) was in good agreement with that obtained in the Cd and Sn microwave-digested serum. A detection limit of 0.14 mg l-1 (3s- Cadmium and Sn are present at a low concentration criterion) was obtained for the serum. (<1 mg l-1). For the determination of Cd and Sn, the ion signals of 111Cd+ and 120Sn+ were monitored. The Cd result Si (see Table 5) was in excellent agreement with the certified range.6 The Sn value obtained (see Table 6) confirmed the Few papers on the determination of Si in serum or blood have lowest literature value of 691 ng l-1 with a standard deviation been published.19 Silicon is one of the most abundant elements of 127 ng l-1 (determined with ICP-MS)17 but had a much in the lithosphere19 and is omnipresent; consequently the risk better precision.Since no spectral interferences are involved of contamination is not fictitious.In addition, the nebulizer and concentrations are not too low, Cd has already previously and spraychamber used in this work were made out of borosilbeen determined via quadrupole ICP-MS.8 icate glass and hence also contributed to the high blank levels. The concentration in the serum reference material analysed in this work may be higher than the actual concentration in Al normal serum. Indeed, the blood used to prepare the serum reference material had been collected into quartz vessels,6 This certified element is mono-isotopic (27Al) and resolution settings of 1454 and 919 are theoretically sucient to separate which undoubtedly provoked a certain degree of contamination. the 27Al+ signal from that of the most important interfering species 12C15N+ and 12C14N1H+,18 respectively.The major Silicon determinations were carried out via the major 28Si isotope and a resolving power of 3000 was used, which is problem to deal with is the high blank, originating from reagents and airborne dust particles19 and yielding an Al largely sucient to avoid spectral interferences from 14N2+ and 12C16O+.In Fig. 2, spectra of the blank, the standard concentration of the same order of magnitude as the five-fold diluted samples. This explains the relatively high standard (20 mgl-1) and of five-fold diluted serum are shown. As can be seen, the 14N2+ and 12C16O+ peaks are clearly separated from deviation. Nevertheless, the result of 1.5 mg l-1 was in good agreement with the certified range.6 Since C is responsible for the Si ion signal.From the blank spectrum, the high degree of Si contamination can be estimated. In Table 3, blank values the major spectral interferences (see Table 4) in quadrupole ICP-MS, this determination was also carried out at low for Si are compared with those in undiluted human serum. The Si blank values are as high as 35% of the concentration resolution in order to check whether the digestion removed C suciently, a partial contribution to which may come from levels in five-fold diluted samples, which explains the relatively high standard deviation.A value of 175 mg l-1 with a standard dissolved CO2 in the solution. The result obtained (6.1 mg l-1) is four times higher than the one obtained with higher reso- deviation of about 10%, has been determined in the reference 936 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12CONCLUSION The double focusing ICP-MS instrument used allowed precise and accurate determination of the certified elements Al, Mn, Fe, Cd, Cu, Zn and Cd in the Human Serum Reference Material.In addition, Si, P, S, Ti, Ag and U were also determined, at a concentration as low as 770 pg l-1 and with a standard deviation of 140 pg l-1 for U. Reconstitution followed by dilution, as sample preparation, combined with flow injection as an introduction technique seemed to be the most successful approach for determination of ultratrace elements in this type of matrix.A resolution setting of 3000 was found to be sucient to eliminate all spectral interferences encountered. Fig. 2 Determination of silicon in human serum at a resolution As illustrated in Table 4, detection limits for many elements, setting of 3000. especially in matrices such as serum, are restricted by blank levels. Hence, additional research will be necessary to reduce these blank levels further, as was illustrated by the biased result for Cr.material, which is comparable to the result of 170 mg l-1 in In the near future, in addition to U, the presence of rare plasma (for 15 normal individuals) from Gitelman and earth elements such as Y, Ce, Lu and Th will be examined in Alderman14 and the values reported by Huang.20 the reference material. Ti J.R. is a Research Assistant and F.V. is a Senior Research Assistant of the Fund for Scientific Research, Flanders. Very little has been published on the Ti levels in serum or blood.21,22 The interest in this element, however, has increased as a result of the increased use of prosthetic devices made out REFERENCES of Ti and the awareness that metallic total joint replacement 1 Alves, L. C., Wiederin, D.R., and Houk, R. S., Anal. Chem., 1992, devices could interact with the surrounding body fluids and 64, 1164. tissues, as was confirmed by Skipor et al.21 2 Gregoire, D. C., J. Anal. At. Spectrom., 1988, 3, 309.Titanium has five isotopes: 46Ti (8.0%), 47Ti (7.3%), 48Ti 3 Tsukahara, R., and Kubota, M., Spectrochim. Acta, Part B, 1990, (73.8%), 49Ti (5.5%) and 50Ti (5.4%). Since the Ca concen- 45, 779. 4 Carey, J. M., Evans, E. H., Caruso, J. A., and Shen, W.-L., tration in the Human Serum Reference Material is about Spectrochim Acta, Part B, 1991, 46, 1711. 90 mg l-1,23 Ti determination via the most abundant isotope 5 Carey, J. M., and Caruso, J. A., Crit. Rev. Anal.Chem., 1992, 48Ti is excluded because of an overwhelming isobaric inter- 23, 397. ference by 48Ca (0.2%). In contrast to ion signals from most 6 Versieck, J., Vanballenberghe, L., De Kesel, A., Hoste, J.,Wallaeys, of the interferences caused by polyatomic ions, isobaric ion B., Vandenhaute, J., Baeck, N., Steyaert, H., Byrne, A. R., and signals cannot be separated from one another with the double Sunderman, F. W. Jr., Anal. Chim. Acta, 1988, 204, 63. 7 Moens, L., Vanhaecke, F., Riondato, J., and Dams, R., J.Anal. focusing mass spectrometer currently used in ICP-MS since At. Spectrom., 1995, 10, 569. the required resolution greatly exceeds 10 000. The second 8 Vandecasteele, C., Vanhoe, H., and Dams, R., J. Anal. At. isotope 46Ti is also excluded for determination due to the Spectrom., 1993, 8, 781. presence of 46Ca (0.03%). Finally, overlap with the 50V+ and 9 Vanhaecke, F., Riondato, J., Moens, L., and Dams, R., Fresenius’ 50Cr+ ion signals made determination using the 50Ti isotope J.Anal. Chem., 1996, 355, 397. impossible. The remaining isotopes, 47Ti and 49Ti, are suitable 10 Giessmann, U., and Greb, U., Fresenius’ J. Anal. Chem., 1994, 350, 186. for measurements. Many polyatomic ions such as 31P16O+ 11 Feldmann, I., Tittes, W., Jakubowski, N., Stuewer, D., and and 12C35Cl+ at m/z=47 and 35Cl14N+, 32S16O1H+ and Giessmann, U., J. Anal. At. Spectrom., 1994, 9, 1007. 12C37Cl+ at m/z=49, however, hamper accurate determina- 12 Moens, L., Verrept, P., Dams, R., Greb, U., Jung, G., and Laser, tions of Ti at R=300 and consequently a resolution of B., J.Anal. At. Spectrom., 1994, 9, 1075. 3000 is required. The low concentrations demand extreme 13 Xilei, L., Van Renterghem, D., Cornelis, R., and Mees L., Anal. care in order to avoid contamination. A detection limit of Chim. Acta, 1988, 211, 231. 14 Gitelman, H. J., and Alderman, R., J. Anal. At. Spectrom., 1990, 0.008 mg l-1 has been reached in human serum and in the 5, 687. reference material a value of 1.1 mg l-1 with a standard devi- 15 Vanhoe, H., Vandecasteele, C., Versieck, J., and Dams, R., ation of 15% has been obtained using the 49Ti isotope. This Mikrochim. Acta, 1989, III, 373. value was confirmed by the result obtained using the 47Ti 16 Yu, L., Koirtyohann, S. R., Rueppel, M. L., Skipor, A. K., and isotope: 0.99 (0.24) mg l-1. Cantone et al. reported a value of Jacobs, J. J., J. Anal. At. Spectrom., 1997, 12, 69. 90 mg l-1 (standard deviation 10 mg l-1) in the serum of five 17 Baumann, H., 1989, personal communication. 18 Reed, N. M., Cairns, R. O., Hutton, R. C., and Takaku, Y., healthy individuals22 but this seems improbably high.17 Skipor J. Anal. At. Spectrom., 1994, 9, 881. et al.21 found a value of 3.3 mg l-1 in serum with a method 19 Versieck, J., and Cornelis, R., T race Elements in Human Plasma detection limit (MDL) of 2.1 mg l-1 by ETAAS. Recently, or Serum, CRC Press, Bacon Raton, FL, 1989. values of 1.3±0.3 and 0.7±0.1 mg l-1 were quoted by Yu et al. 20 Huang, Z., Spectrochim. Acta, Part B, 1995, 50, 1383. using ETV-ICP-MS as technique for which a detection limit 21 Skipor, A. K., Jacobs, J. J., Schvocky, J., Black, J., and Galante, of 0.4 mg l-1 was obtained.16 J. O., At. Spectrosc., 1994, 15, 131. 22 Cantone, M. C., Molho, N., and Pirola, L., J. Radioanal. Nucl. Chem., 1984, 91/1, 197. U 23 Vanhoe, H., Ph. D. Thesis, University of Gent, 1992. No values of U in serum have been reported up till now. Paper 7/01652B Measurements were made at low resolution since no spectral ReceivedMarch 10, 1997 interferences are to be found at an m/z of 238. As was to be Accepted June 13, 1997 expected, an extremely low value (0.8 ng l-1) was found. Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 937