Determination of trace metals in size fractionated particles from arctic air by electrothermal vaporization inductively coupled plasma mass spectrometry{ Christian Lu»dke,*a Erwin Hoffmann,a Jochen Skolea and Michael Kriewsb aInstitut fu»r Spektrochemie und Angewandte Spektroskopie, Institutsteil Berlin, Albert-Einstein-Strasse 9, 12489 Berlin-Adlershof, Germany bAlfred-Wegener-Institut fu»r Polar- und Meeresforschung, Am Handelshafen 12, 27570 Bremerhaven, Germany Received 12th May 1999, Accepted 27th August 1999 Studies of element composition in small atmospheric particles aid the clariÆcation of processes such as longrange transport, deposition and transformation of particles and quantiÆcation of emission from natural and anthropogenic sources.For this purpose, a highly sensitive method was developed for the trace analysis of atmospheric particles. The particles were sampled and separated according to size, directly on separate small graphite discs arranged behind the jet-nozzles of an eight-stage cascade impactor.To determine the elemental composition of the particles, the ETV-ICP-MS technique was applied. In an appropriately sealed electrothermal vaporizer, linked to an inductively coupled plasma mass spectrometer, the targets were heated and the sample vapour was swept by argon into the plasma. The system described was used for the analysis of long-range transported particles from Arctic air sampled at the German Arctic research station at Spitsbergen, Norway, in spring 1998.For the elements Mn, Fe, Co, Ni, Ag, Cd, Sn, Sb and Pb the trace element content per cubic metre of air was measured as a function of the aerodynamic particle diameter. Air masses of different origin cause characteristic particle distributions at low changes in total dust burden. The relative detection limits for the elements measured in an air volume of 0.275 m3 were determined to be within 0.3±10 pg m23; the overall analytical precision was around 20% for all trace metals.Introduction Both natural and anthropogenic sources emit particles into the atmosphere but the predominant part of metallic trace elements is of anthropogenic origin.1 Potentially toxic metals such as Pb, Cd, Ni, Sb and Tl associated with atmospheric particulate matter are emitted into the atmosphere mostly by burning of fossil fuels, metallurgical plants and smelters and increasingly by waste incineration.2 Ambient atmospheric particles from anthropogenic and natural sources were found to be predominantly oxides, sulfates, chlorides and silicates such as feldspars and black mica.3,4 The greater part of particulates emitted is deposited about the vicinity of the source but, depending on the meteorological conditions and particle size, they may be subjected to long-range transport and will reach remote areas far away from the source regions.Especially in the northern hemisphere where anthropogenic sources are concentrated in Europe, North America and Siberia, the natural cycles are strongly affected by anthropogenic emissions.Well known is the signiÆcant seasonal variation of trace metal levels in the Arctic, where meteorological conditions strongly favour a winter/spring burden. During summer the Arctic is cut off from the pollution sources but in winter and spring the polar front extends to the south over the source areas.5 This leads to a transport of aerosol, at low altitudes, into the Arctic region and causes pollution by anthropogenic mid-latitude emission sources.In this way the `Arctic haze' so termed by Mitchell,6 a dust layer at low altitudes with turbidity as usually encountered in industrial regions, is generated. The examination of this phenomenon, increasing emissions of air polluting substances and the expanding mineral oil extraction industry in the Arctic have led to a need for trace element determinations. In recent studies7±9 on ground-level distributions of particles in arctic aerosol, Æltration and impaction techniques were used for sampling.Hundreds of cubic metres of air were sucked through Ælters for subsequent analysis by particle-induced X-ray emission (PIXE) and instrumental neutron activation analysis (INAA). As, Mn, Sb, Se, V and Zn were measured by INAA, Cr, Ni and Pb by PIXE and Cd in a nitric acid extract by electrothermal atomic absorption spectrometry (ETAAS). The concentrations obtained were in the ng m23 range with an error between 5 and 20%, depending on the element and method.8 To collect size fractionated aerosol samples, a Batelle-type cascade impactor was operated for 5 d and the fractionated aerosol samples were analysed by PIXE.10 The analytical methods mentioned above require time consuming sampling of large air volumes for the determination of the expected very low trace element levels.However, when the time-scale for changes in atmospheric conditions limits the sampling time, a powerful analytical technique is required, combined with a highly efÆcient sampling system if possible.Inductively coupled plasma mass spectrometry (ICP-MS) offers new opportunities for the multi-element determination of trace metals and in combination with electrothermal vaporization (ETV) also for the analysis of solid samples. The Ærst reference to the use of ETV with ICP-MS was by Gray and Date in 1983.11 Several workers have applied this method in practical situations, e.g., for the determination of As and Se in solid materials,12 trace metals in sea-water13 and trace metals in arctic snow.14,15 Coupling ICP-MS with ETV, in which samples are volatilized from a graphite furnace and transported as a dry aerosol into the plasma, allows the direct analysis of solid samples.16±18 Following this idea, a sampling system was {Presented at the European Winter Conference on Plasma Spectrochemistry, Pau, France, January 10±15, 1999.Dedicated to Prof. Dr. Dieter Klockow on the occasion of his 65th birthday. J. Anal. At. Spectrom., 1999, 14, 1685±1690 1685 This Journal is # The Royal Society of Chemistry 1999developed in which airborne particles were impacted on graphite targets and subsequently analyzed by ETV-ICP-MS. The primary aim of this study was to evaluate the newly developed method for sampling particles fractionated according to size, by coupling with the powerful multi-element analysis technique ETV-ICP-MS.As the object of this study the characterization of long-range transported Arctic air particles was chosen. For this study, aerosol sampling was performed at the German Arctic research station in Ny- A lesund (78.9�N, 11.9�E) on Spitsbergen, Norway, in a measurement campaign between March 10 and 30, 1998. Experimental Sample collection Many different sampling devices have been developed for the collection of airborne particulates.19 The method used in our studies was inertial deposition and size fractionation by a cascade impactor.The operating principle of a cascade impactor is well described in the literature.20,21 We used an eight-stage multijet cascade impactor made by Stro» hlein Instruments (Kaarst, Germany). It separates particles at a Øow rate of 36.7 l min21 in eight size classes with cut-off diameters between 0.35 and 16.5 mm. The cut-off particle diameter corresponds to a 50% collection efÆciency and depends not only on the particle size but also on the density and the particle shape.The calculated mass, assuming a spherical particle of density 2 g cm23, ranges between 4.5610214 g (0.35 mm diameter) and 4.761029 g (16.5 mm diameter). Usually, the separation characteristic for cascade impactors is given for spherical particles of unit density. The diameter of actual particles differing from this ideal is expressed as the aerodynamic diameter, dae. Owing to the gas velocity and the resulting Stokes number at a sample volume of 2.2 m3 h21 for our experimental setup, particles larger than 16.5 mm were not collected.Particles smaller than 0.35 mm were collected on a backup Ælter made of quartz- or glass-Æbre. The detailed sampling procedure has been described in previous publications22,23 and only a short summary is given here. Tmpaction plates were newly designed, made of poly(methyl methacrylate) (PMMA), and covered with separate porous graphite impaction targets in such a way that one target is placed under one nozzle.The graphite targets, discs of 6 mm diameter with a central impaction area of 4 mm diameter, 2 mm thick, were manufactured in-house from a rod of pure graphite (RW 003, Ringsdorff-Werke, Bonn, Germany). The chemical characterization of the collected particles requires their sampling on a material which does not interfere with the subsequent analytical procedure. The rough surface of the graphite discs yield a sufÆcient sampling efÆciency and a minimized loss of particles due to elastic collisions.For the measurement campaign, sets of impaction plates were prepared and each plate was equipped with six targets. Well sealed containers as shown in Fig. 1 were used for the transport of the impaction plates. At the scene of sampling the set of impaction plates in the impactor was replaced daily with a new one. To avoid Æeld blanks, replacing was done in a clean bench (US class 100) at the sampling station.During the sampling time of 23 h per day a mean volume of 275 l per target was sampled. Particles impacted on the graphite targets were subsequently analysed in our laboratory in Berlin. ETV-ICP-MS measurements For the multi-element analysis, a Perkin-Elmer SCIEX (Thornhill, ON, Canada) Elan 5000 ICP mass spectrometer was used. As the furnace for the ETV a modiÆed FANES source24,25 manufactured by BBPT Gesellschaft fu» r Physikalisch- technischen Gera»tebau (Berlin, Germany) was used.The anode chamber was replaced by an adapter which accommodates the injection tube of the ICP torch. In this way the electrothermal vaporizer is connected to the torch as closely as possible, avoiding transport losses and non-linear calibration curves.26,27 A scheme of the experimental setup is given in Fig. 2. Two gases are of importance for the performance of the furnace: the stabilization gas and the transport gas.The stabilization gas is an additional gas Øow of 1.2 l h21 argon introduced via the PTFE adapter between the ICP and the furnace. It replaces a mechanical shutter and prevents the Øow of air into the plasma, which causes it to break down, when the lid is lifted or the pivoted arm of the furnace is open. When drying a liquid sample dosed on the target, the part of the stabilization gas streaming through the tube removes the water and other vapours via the dosing hole, supported by the transport gas from the opposite end of the tube.The argon transport gas with a Øow rate of 1 l min21 transports the vaporized material into the ICP. The normalized height of the measured signal for the elements studied as a function of the transport gas Øow through the furnace is given in Fig. 3. The instrumental operating conditions for the Elan 5000 are given in Table 1. Combined with ETV, the Elan instrument was operated in the graphic mode. This application allows one to monitor the intensities of given isotopes as a function of time.The total measuring time was 20 s per heating cycle, triggered by a read signal of the ETV power supply. The intensity±time Æles of each run were evaluated by an in-house developed computer program which permits settings of individual integration limits for each element. The integration times Fig. 1 Set of modiÆed impaction plates. Fig. 2 Scheme of coupling ETV to ICP-MS. 1686 J. Anal. At. Spectrom., 1999, 14, 1685±1690vary between 7 and 10 s and the half-width of the peaks between 1 s (Ag, Cd, Pb) and 3 s (Fe, Ni, Co).Procedure To analyse the impacted particles, the loaded graphite targets were transferred one after the other into a two-part graphite tube container and Æxed there in two slits like a platform. Moving the pivoted arm of the ETV opens the furnace for changing the tube. The two-part graphite tube container which houses the target for ETV was manufactured and coated with pyrolytic graphite to speciÆcation by Ringsdorff-Werke. The closed tube is 28 mm long and has a 6 mm id and 10 mm od in the central part.To minimize the uncertainties in ETV solid sampling measurements, a thoroughly developed concept of the following steps was applied: (1) the ion optics and the plasma position were optimized by nebulization of aqueous standards prior to changeover to ETV; (2) the ETV was mounted behind the injection tube of the ICP torch as closely as possible; (3) the ETV and the power supply were calibrated using a fast pyrometer to operate under computer control at true stabilized temperatures; (4) the graphite targets were cleaned by repeated heating to 2850 �C, according the clean-up programme in Table 2, until stable background signals were obtained; and (5) particle-loaded targets were heated at 2750 �C for 6 s so that more than 90% of the whole signal was measured in the Ærst heating cycle.Mostly in the third consecutive heating cycle the background signal, as determined in advance, will be obtained.For calibration of measurements with the graphite disc, acidic multi-element standard solutions of 0.5, 2.5, 5, 25, 50, 100 and 200 mg l21 were used. The calibration standards for Mn, Fe, Co, Ni, Ag, Cd, Tl and Pb were freshly prepared in 0.029 mol dm21 HNO3 by stepwise dilution of 1 g l21 stock standard solutions from Merck (Darmstadt, Germany). Standards for Sb and Sn were analogously prepared in 0.024 mol l21 HCl.An aliquot of 10 ml of calibration solution was delivered manually to the cleaned disc inside the tube using a micropipette. The heating programme, as given in Table 2, was then started with the lid open at the sample injection port; it was automatically sealed after removing all vapours in the charring step. For all elements studied, linear calibration curves were obtained. The linear regression coefÆcient (r2), the concentration range and the sensitivity for each analyte isotope are given in Table 3.Differences in sensitivity reported in Table 3 are caused by compromise conditions for the furnace heating programme which was needed to determine volatile and nonvolatile elements in a single run. The sensitivity in ETV-ICPMS depends on ionisation energy but much more on element speciÆc properties such as number of isotopes, multiple charging and time dependent thermochemical processes of volatilization, dissociation and ionisation.Detection limits for each element based on 3s blank values are given in Table 4. The higher LODs for Fe and Ni are caused by incomplete cleaning after manufacture resulting in higher blank values for both of these elements. The accuracy of calibration was veriÆed by comparison with other methods and from isoformation by gas-phase digestion described previously.28 The reproducibility of repeated injections of reference solution is near 8% (relative standard deviation).Table 3 presents an example for 10 repeated injections of the 50 mg l21 standard solution. The stability of the instrument was checked daily by measuring the 50 mg l21 calibration point. The relative deviation of the mean intensity for 50 mg l21 varies between 12 and 20% over 10 d for the analyte ions concerned (see Table 3). A correction for residual instabilities of the ETV-ICP-MS Table 1 Instrumental operating conditions for ICP-MS ICP– Rf power 1000 W Plasma gas 14.8 l min21 argon Auxiliary gas 0.8 l min21 argon Nebulizer gas~transport gas 1.0 l min21 argon Sampler and skimmer cone Pt MS– Application Graphics (displays the evolution of a transient signal during an analysis) Number of replicates 80 Dwell time 20 ms Total measuring time 20 s Isotopes measured 55Mn, 57Fe, 59Co, 60Ni, 107Ag, 111Cd, 118Sn, 121Sb, 208Pb Scan mode Peak hopping Signal evaluation Peak integral (calculation performed at external PC) Fig. 3 Normalized signal as a function of the transport gas Øow averaged over all of the isotopes measured.Table 2 Operating programme for ETV Step Temperature/�C Ramp rate/�C s21 Hold time/s Argon transport gas/l min21 Argon stabilization gas/l min21 0 60 20 5 1 1.2 1 150 4 20 1 1.2 2 55a 1 1.2 3 2750 590 4b 1 0c 4 2750 0 6 1 0c 5 15 0 15 0 1.2c The clean-up programme was the same as the operating programme without steps 0 and 1 and with 2850 �C in steps 3 and 4. aData aquisition triggered 12 s after the beginning of the step.bRamp time. cInjection port lid closed. J. Anal. At. Spectrom., 1999, 14, 1685±1690 1687combination, despite the application of a true temperature controlled furnace, is possible by using a well suited internal standard. Measurements were made here without internal standardisation because the precision is sufÆcient to ensure that the differences in the sampled air volumes were determined exactly. Results and discussion Detailed measurements of airborne trace element concentrations at Ny-A lesund were carried out in March 1998.In the following a selected period of four successive days from March 25 to 28 is considered. During this period only weak air movements occurred. The air mass origin was concluded from 5 d backward trajectory analysis (850 mbar, isobaric) calculated by the German Weather Service (DWD) and the Alfred- Wegener-Institute (Potsdam, Germany). The pathways of the air packages which were measured are indicated in Fig. 4. The air masses reaching the station on March 25 (open circles) and March 28 (closed circles) had passed mainly over ice covered areas of the Arctic Ocean. Only on March 26 (solid line) and March 27 (broken line) did the air pass the Siberian coastline and the northernmost part of Siberia. On each day one impactor was operated equipped with eight impaction stages, corresponding to the eight separate size classes (0.35±16.5 mm dae), covered with six graphite targets per stage.From the determined element concentration of each target, the mean stage concentration and standard deviation (n~6) were calculated. Summation over all stages gives a bulk element concentration per day with a standard deviation calculated according to error propagation. Table 5 shows the results. It is obvious that the concentrations show a variability by a factor of 2±3. This cannot be explained by the inhomogeneity of the sampled aerosols. Aerosol sampling on Ælters with three parallel samplers showed a variability between the samplers and the distribution on the Ælters itself in a range of about 5±25% only, depending on the elements which were measured.29,30 The concentrations shown in Table 5 are in very good agreement with data obtained on non-size separated Ælter samples during the same period on a weekly sampling basis.31 Fig. 5 presents the histograms of the percentage related to the bulk element concentration over the eight size classes, (a) for Mn, Fe and Ni and (b) for Ag, Cd and Pb.The elements Co, Sb and Sn, also evaluated, are not shown in Fig. 5 since they follow nearly the same curves. However, as can be seen from Fig. 5, there are signiÆcant differences in elemental size distributions depending on air mass history. The histograms show a signiÆcant change in particle size distribution depending of the path of the air package which was sampled. Fairly clean air from the Arctic Ocean reached Ny A lesund on March 25 and 26 with a typical size distribution for unpolluted longrange transported aerosols.The air masses passing Siberia and the Siberian coast line (March 26 and 27) showed a different size distribution with a shift of the concentration maxima to 3.45 mm aerodynamic diameter on March 27. Table 3 Calibration data Analyte Concentration range/pg Linear regression coefÆcient Sensitivity/ counts pg21 Background counts (mean°s, n~18) Reproducibilitya (100s/mean, n~10) (%) Stabilityb (100s10/mean10) (%) 55Mn 5±2000 0.9958 5220 1418°194 10.0 16 57Fe 50±2000 0.9904 99 4572°835 8.3 18 59Co 5±2000 0.9916 2810 3179°788 7.3 15 60Ni 25±2000 0.9921 538 7719°1525 5.6 14 107Ag 5±2000 0.9924 990 120°28 9.6 12 111Cd 5±2000 0.9919 471 773°99 8.4 17 118Sn 5±2000 0.9998 269 2568°184 – – 121Sb 5±2000 0.9998 93 678°104 – – 208Pb 5±2000 0.9982 1713 1321°195 7.5 20 an repeated injections of 10 ml of the 50 mg l21 standard solution. bRelative deviation of the mean of Æve repeated injections on 10 successive days.Table 4 Limits of detection (LOD) (3s blank, n~18) Analyte Absolute LOD/pg Relative LODa/pg m23 55Mn 1 4 57Fe 25 90 59Co 0.8 3 60Ni 8 30 107Ag 0.08 0.3 111Cd 0.6 2 118Sn 2 7 121Sb 3 10 208Pb 0.3 1 aRelative LOD/pg m23~ absolute LOD=pg 0:275 m3 sample volume Fig. 4 Calculated wind trajectories, March 25 (open circles), March 26 (solid line), March 27 (broken line) and March 28 (closed circles). Table 5 Bulk element concentrations/ng m23 Analyte March 25 March 26 March 27 March 28 Mn 6.8°1.2 2.5°0.2 4.0°0.4 6.5°0.3 Fe 335°46 87°7 170°15 318°12 Ni 58°10 24°7 45°8 25°7 Co 1.8°0.8 0.83°0.24 0.39°0.07 0.38°0.07 Sn 9.5°1.0 8.1°2.2 9.0°0.4 14.8°2.2 Sb 8.0°0.5 8.1°0.7 9.8°0.4 11.8°0.6 Ag 0.2°0.03 0.23°0.09 0.09°0.01 0.08°0.01 Cd 0.47°0.06 9.1°3.7 0.37°0.11 0.36°0.09 Pb 12.4°1.4 14.3°3.4 16.1°2.7 12.9°2.2 1688 J.Anal. At. Spectrom., 1999, 14, 1685±1690The air mass trajectories indicate transport of air polluting substances from northern Russia as the main source region.Sites of heavy industry are located there (Pechora basin, Norilsk area) such as mining industry and metal smelters.32 The estimated high contents of Ni, Fe, Cd, Pb and Sb point to emissions from the large copper±nickel smelters of the Norilsk complex. The very similar shape of the curves for all elements, seen in Fig. 6, indicates a common source. From the results dicussed above, it can be seen that the combination of a welldesigned sampling strategy with subsequent high performance element analyses and backward trajectory calculations leads to detailed information about atmospheric transport processes of pollutants from highly industrialised source regions to remote areas such as the high Arctic.Conclusion and outlook It has been demonstrated that trace elements in size classiÆed atmospheric particles can be determined successfully by ETVICP- MS. The impaction of particles on graphite targets permits their direct analysis without any sample preparation and very good contamination control.The detection power is high enough for all elements studied to allow sampling times of only about 2 h. This newly designed sampling technique in combination with a subsequent high performance analysis technique leads to a much better time resolution for studying short time atmospheric processes, in comparison with wellestablished methods, in a remote area such as polar regions.A remarkable result of this study is the more or less strong variation in the particle size distribution, depending on the meteorological conditions, at a relatively low variation of the bulk element concentration. Further improvements in performance capability can be expected by using truly simultaneous mass spectrometers, e.g., time-of-Øight instruments. In this way, the number of simultaneously measurable isotopes is not limited when fast transient signals are measured.Such instrumentation also promises the availability of accurate isotope ratio information, which can be used for source identiÆcation. Fig. 5 Percentage content of (a) Mn, Fe and Ni and (b) Ag, Cd and Pb versus the aerodynamic diameter. The error bars indicate the standard deviation calculated for six targets per size class. The 100% content of each element between 0.35 and 16.5 mm aerodynamic diameter is given in Table 5. Fig. 6 Measured element content of air versus the aerodynamic diameter, March 27.The error bars indicate an uncertainty of 20%. J. Anal. At. Spectrom., 1999, 14, 1685±1690 1689Acknowledgements The �¡nancial support of the Senatsverwaltung fu¡í r Wissenschaft, Forschung und Kultur des Landes Berlin and the Bundesministerium fu¡í r Bildung und Forschung is gratefully acknowledged.erences 1 J. O. Nriagu, Nature (London), 1989, 338, 47. 2 M. V. Johnston and A. S. Wexler, Anal. Chem.., 1995, 67, 721A. 3 J.Mu¡í ller, J. Aerosol Sci., 1998, 29 (Suppl. 1), S219. 4 J. Kasparian, E. Frejafon, P. Rambaldi, J. Yu, B. Vezin, J. P. Wolf, P. Bitter and P. Viscard, Atmos. Environ., 1998, 32(17), 2957. 5 L. A. Barrie, Atmos. Environ., 1986, 20, 643. 6 M. Mitchell, J. Atmos. Terr. Phys., Suppl., 1956, 195. 7 J. Heintzenberg, H. C. Hansson and H. Lannefors, Tellus, 1981, 33, 162. 8 J. M. Pacyna, B. Ottar, U. Tomza and W. Maenhaut, Atmos. Environ., 1985, 19, 857. 9 B. Ottar and J.M. Pacyna, Geophys. Res. Lett., 1984, 11, 441. 10 B. Ottar, Technical Report 30/86: Air Pollutants in the Arctic, Norwegian Institut for Air Research, Lillestr¢�m, 1986. 11 A. L. Gray and A. R. Date, Analyst, 1983, 108, 1033. 12 S. Boonen, F. Vanhaecke, L. Moens and R. Dams, Spectrochim. Acta, Part B, 1996, 51, 271. 13 G. Chapple and J. P. Byrne, J. Anal. At. Spectrom., 1996, 11, 549. 14 R. E. Sturgeon, S.N. Willie, J. Zheng, A. Kudo and D. C. Gregoire, J. Anal. At. Spectrom., 1993, 8, 1053. 15 M. Kriews and O. Schrems, J. Aerosol Sci., 1998, 29, 735. 16 D. C. Gregoire,M. Lamoureux, C. L. Chakrabarti, S. Al-Maawali and J. P. Byrne, J. Anal. At. Spectrom., 1992, 7, 79. 17 C. M. Sparks, J. Holcombe and T. L. Pinkston, Spectrochim. Acta, Part B, 1993, 48, 1607. 18 M. M. Lamoureux, D. C. Gregoire, C. L. Chakrabarti and D. M. Goltz, Anal. Chem., 1994, 66, 3208¡¾3222. 19 D. Klockow, Fresenius' Z. Anal. Chem., 1987, 326, 5. 20 D. Hochrainer, in Analysis of Airborne Particles by Physical Methods, ed. H. Malissa and J. W. Robinson, CRC Press, West Palm Beach, FL, 1978, p. 7. 21 V. A. Marple and K. Willeke, in Inertial Impactors in Aerosol Measurements, ed. D. A. Lundgren, F. S. Harris, Jr., W. H. Marlow, M. Lippmann, W.-E. Clark and M. D. Durham, University of Florida Press, Gainsville, FL, 1979, p. 90. 22 C. Lu¡ídke, E. Hoffmann, J. Skole and S. Artelt, Fresenius' J. Anal. Chem., 1996, 355, 261. 23 C. Lu¡í dke, E. Hoffmann and J. Skole, Fresenius' J. Anal. Chem., 1997, 359, 399. 24 H. Falk, E. Hoffmann and C. Lu¡í dke, Prog. Anal. Spectrosc., 1988, 11, 417. 25 D. C. Baxter, R. Nichol, D. Littlejohn, C. Lu¡í dke, J. Skole and E. Hoffmann, J. Anal. At. Spectrom., 1992, 7, 727. 26 R. D. Edinger and S. A. Beres, Spectrochim. Acta, Part B, 1992, 47, 907. 27 T. Kantor, Spectrochim. Acta, Part B, 1988, 43, 1299. 28 C. Lu¡ídke, E. Hoffmann and J. Skole, J. Anal. At. Spectrom., 1994, 9, 685. 29 M. Kriews, W. Dannecker, K. Naumann and U. Wa¡í tjen, J. Aerosol Sci., 1988, 7, 1295. 30 U. Wa¡í tjen, M. Kriews and W. Dannecker, Nucl. Instrum. Methods Phys. Res., 1990, 49, 360. 31 M. Kriews and O. Schrems, J. Aerosol Sci., 1998, 29 (Suppl. 1), S685. 32 C. Reimann, H. Nikavaara, P. de Caritat, M. A¡í yra¡ís, V. A. Chekushin and T. E. Finne, in Heavy Metals in the Environment, ed. R.-D. Wilken, U. Fo¡í rstner and A. Kno¡í chel, CEP Consultants, Edinburgh, 1995, vol. 1, pp. 84¡¾87. Paper 9/03815I 1690 J. Anal. At. Spectrom., 1999, 14, 1685¡¾16