Anal. Chem. 1999, 71, 4125-4133
Determination of Rh, Pd, and Pt in Polar and Alpine Snow and Ice by Double-Focusing ICPMS with Microconcentric Nebulization Carlo Barbante,*,†,‡ Giulio Cozzi,† Gabriele Capodaglio,†,‡ Katja Van de Velde,§ Christophe Ferrari,§,| Audrey Veysseyre,§ Claude F. Boutron,§,⊥ Giuseppe Scarponi,+,# and Paolo Cescon†,‡
Dipartimento di Scienze Ambientali, Universita` Ca’ Foscari di Venezia, Dorsoduro 2137, I-30123 Venezia, Italy, Laboratoire de Glaciologie et de Ge´ ophysique de l’Environnement du CNRS, 54, rue Molie` re, Domaine Universitaire, B.P. 96, 38402 Saint Martin d’He` res Cedex, France, and Istituto di Scienze del Mare, Universita` di Ancona, via Brecce Bianche, I-60131 Ancona, Italy
High sensitivity coupled with capability of working with low amounts of sample is an important requirement for instrumental analytical chemistry when scientifically valuable samples, both
extremely dilute and limited in quantity, need to be analyzed. The determination of trace elements in ice cores collected in polar and alpine sampling sites represent a fascinating example of this challenging field of chemical analysis, where only a few milliliters of samples are generally available for the analysis.1 Platinum group elements (PGE), particularly rhodium, palladium, and platinum, have recently come to be mobilized into the environment by human activity, mainly by the exhaust gases of cars equipped with catalytic converters.2,3 Because of the increasing concern about these elements as potential risks for human health, owing to possible direct contact by inhalation of dust, many efforts have been made in the past two decades to quantify them in different environmental matrixes.4-6 Research has also focused on reconstruction of recent historical trends by studying different archives such as on-shore marine sediments 7 and grass samples collected yearly near a road carrying considerable traffic.5 A good possibility to reconstruct the past variations and recent changes in the chemical composition of the atmosphere at a global scale is given by the study of snow and ice collected in polar or high-elevation alpine sites.1 Since the development of the inductively coupled plasma mass spectrometry (ICPMS), the potential of this new technique for the determination of PGE in biological, geological, and environmental samples has been clear.8,9 Very high sensitivity (detection limits at a subpicograms per gram level in most environmental matrixes, 1 pg g-1 ) 10-12 g g-1), multielement capability, low sample consumption, and high sample throughput (∼40 samples per day), were among the best appreciated features. ICPMS, using a quadrupole mass analyzer, opened the door to the simultaneous determination of several trace elements in snow, such as Co, Mo,
* Corresponding author (e-mail)
[email protected]. † Universita ` Ca’ Foscari di Venezia. ‡ Also at Centro di Studio sulla Chimica e le Tecnologie per l’Ambiente - CNR, Dorsoduro 2137, I-30123 Venezia, Italia. § Laboratoire de Glaciologie et Ge ´ ophysique de l’Environment du CNRS. | Also at Institut des Sciences et Techniques de Grenoble, Universite ´ Joseph Fourier, 28 Avenue Benoıˆt Frachon, B.P. 53, 38041 Grenoble, France. ⊥ Also at Unite ´ s de Formation et de Recherche de Me´canique et de Physique, Universite´ Joseph Fourier de Grenoble (Institut Universitaire de France), Domaine Universitaire, B.P. 68, 38041 Grenoble, France. + Universita ` di Ancona. # Also at Universita ` Ca’ Foscari di Venezia, Dorsoduro 2137, I-30123 Venezia, Italia.
(1) Boutron, C. F. Environ. Rev. 1995, 3, 1-28. (2) Helmers, E. Environ. Sci. Pollut. Res. 1997, 4, 100-103. (3) Barefoot, R. R. Environ. Sci. Technol. 1997, 31, 309-314. (4) Balcerzac, M. Analyst 1997, 122, 67R-74R. (5) Helmers, E.; Mergel, N. Fresenius J. Anal.Chem. 1998, 362, 522-528. (6) Helmers, E.; Schwarzer, M.; Schuster, M. Environ. Sci. Pollut. Res. 1998, 5, 44-50. (7) Lee, D. S. Nature 1983, 305, 47-48. (8) Krachler, M.; Alimonti, A.; Petrucci, F.; Irgolic, K. J.; Forastiere, F.; Caroli, S. Anal. Chim. Acta 1998, 363, 1-10. (9) Lustig, S.; Zang, S.; Michalke, B.; Schramel, P.; Beck, W. Fresenius J. Anal. Chem. 1997, 357, 1157-1163.
The performance of a double-focusing inductively coupled plasma mass spectrometer equipped with a microconcentric nebulizer was investigated for the direct simultaneous determination of Rh, Pd, and Pt in less than 1 mL of melted snow and ice samples originating from remote sites. Ultraclean procedures were adopted in the laboratories and during the pretreatment steps, to avoid possible contamination problems. Spectroscopic and nonspectroscopic interferences affecting the determination of Rh, Pd, and Pt were carefully considered. Detection limits of 0.02, 0.08, and 0.008 pg g-1 for Rh, Pd, and Pt, respectively, were obtained using the following isotopes: 103Rh, 106Pd, and 195Pt. Repeatability of measurements, as RSD, was 27, 28, and 29%, for Rh, Pd, and Pt, respectively. The new method was applied to the analysis of samples coming from Greenland, Antarctica, and the Alps in order to assess the past natural background concentrations and to determine the present level of these polluting substances. The extremely low detection limits allowed the direct analysis of all samples except for two Greenland ice core sections dating from 7260 and 7760 years ago for which a preconcentration step was necessary. Concentration ranges for all snow samples were (in pg g-1) as follows: Rh (0.0005-0.39), Pd (0.01-16.9), and Pt (0.008-2.7). The lowest concentrations were measured in the enriched Greenland ancient ice samples.
10.1021/ac981437g CCC: $18.00 Published on Web 08/24/1999
© 1999 American Chemical Society
Analytical Chemistry, Vol. 71, No. 19, October 1, 1999 4125
Ag, Sb, Mn, Tl, Cu, Ni, Cr, and U.10-12 Nevertheless, to our knowledge only our preliminary work has reported the application of double-focusing ICPMS using a normal concentric nebulizer.13 The use of double-focusing ICPMS is of paramount importance in cases of extremely low analyte concentrations, since its bent geometry keeps the noise at a very low level and guarantees the highest ion transmission. In this paper, we present a new method for the direct simultaneous determination of Rh, Pd, and Pt in snow and ice samples from remote sites by double-focusing ICPMS using a microconcentric nebulizer (MCN). The use of this sample introduction system with respect to the Meinhard nebulizer type A, previously adopted,13 has been implemented to reduce to a minimum the amount of sample consumed, since the volume of sample available in the study of heavy metals in polar and alpine snow and ice is often less than 1 mL. The potential interferences affecting the determination and the analytical capability of the technique are critically assessed with the aim of developing a reliable and efficient analytical procedure to monitor global-scale contamination by PGE. Results of measurements carried out on samples collected in Greenland, Antarctica, and the Alps are presented and a comparison with the previously described method based on a normal concentric nebulizer is provided. EXPERIMENTAL SECTION Laboratories and Materials. All the analytical procedures preceding the instrumental measurements (i.e., cleaning of all plastic itemsslow-density polyethylene (LDPE)sand of sampling and storage bottles, preparation of standard solutions, and decontamination of core sections) were carried out in class 100 warm and cold clean chemistry laboratories.13,14 Ultrapure concentrated HNO3 (70%) supplied by the National Institute of Standards and Technology (NIST, Gaithersburg, MD) was used for the acidification of samples and standards (1:200 diluted) and for the final steps of the cleaning procedures of bottles and other items. Chloroform (Merck, Darmstadt, Germany) and Suprapur grade HNO3 (65%, Merck) were used at the beginning of the cleaning procedures. Ultrapure water was obtained by coupling a Milli-RO with a Milli-Q system (Millipore, Bedford, MA) or by a mixed bed of ion-exchange resins from Maxy (La Garde`, France). Acidified (NIST HNO3, 1:200 diluted) multielement standard solutions of Rh, Pd, and Pt were prepared through successive dilutions with ultrapure water, in precleaned low-density polyethylene bottles, of 1000 mg L-1 ICPMS stock solutions (Janssen Chimica, Geel, Belgium for Rh and Spex Certiprep, Metuchen, NY, for Pd and Pt). The concentrations in the final standard solutions were 0.05, 0.1, 0.2, 0.5, 1.0, and 2.0 pg g-1 for the three elements. Standard solutions were stored frozen until the day of analysis to minimize possible interactions with the walls of the (10) Davies, T. D.; Tranter, M.; Jickells, T. D.; Abrahams, P. W.; Landsberger, S.; Jarvis, K.; Pierce, C. E. Atmos. Environ. 1992, 26A, 95-112. (11) Jickells, T. D.; Davies, T. D.; Tranter, M.; Landsberger, S.; Jarvis, K.; Abrahams, P. W. Atmos. Environ. 1992, 26A, 393-401. (12) Sturgeon, R. E.; Willie, S. N.; Zheng, J.; Kudo, A.; Gre´goire, D. C. J. Anal. At. Spectrom. 1993, 8, 1053-1058. (13) Barbante, C.; Bellomi, T.; Mezzadri, G.; Cescon, P.; Scarponi, G.; Morel, C.; Jay, S.; Van de Velde, K.; Ferrari, C.; Boutron, C. F. J. Anal. At. Spectrom. 1997, 12, 925-931. (14) Boutron, C. F. Fresenius J. Anal. Chem. 1990, 337, 482-491.
4126 Analytical Chemistry, Vol. 71, No. 19, October 1, 1999
bottle. Multielement and single-element standard solutions of ICPMS stock solutions (1000 mg L-1, Merck, and 1000 mg L-1, Spex Certiprep, Metuchen, NY) were used for study on the potential interferences. Samples and Sample Treatments. Snow and ice samples were collected in Greenland, in Antarctica, and in the Alps to have an overview of North and South Hemispheres, together with remote and industrialized areas. Samples covered different time periods, including old Greenland ice samples (thousands of years old), to compare recent snow data with past natural Holocene PGE concentrations. The Greenland sampling site was located at Summit (central Greenland). Selected samples were as follows: (a) two ice core sections of the 3028.8-m European Greenland Ice Core Project (GRIP) deep ice core,15 dating from 7260 and 7760 years ago; (b) a comprehensive set of 22 snow core sections covering the period 1969-1988 collected in 1989 by shallow hand-drilling;16 (c) 68 shallow snow samples, covering the period 1991-1995.13 Antarctic samples were as follows: two surface snow samples collected in 1995 near Dome C (east Antarctica) and on the Mount Melbourne (Victoria Land), respectively; three shallow snow core samples dating back to 1983, 1977, and 1969, respectively, were also collected in 1991 on the Styx Glacier plateau (Victoria Land).17 Alpine sampling sites were located at Doˆme de Gouˆter in the Mont Blanc range and at Colle Gnifetti in the Monte Rosa massif. Selected samples were as follows: (a) Doˆme de Gouˆter; 74 sections of an ice and snow core covering a period from about 1778 to 1991;18 (b) Colle Gnifetti; 25 sections of a snow core covering a period from 1973 to 1995. The ages of the snow/ice core samples were estimated by chemical profiles measured in different subaliquots of the same samples13,16,19 or roughly estimated (Alpine samples) from models based on the average accumulation rates of the sampling sites and from the densities.20 Samples were transported frozen from the field to the laboratory, where they were treated, melted, divided into different aliquots, acidified (NIST HNO3, 1:200 diluted), and then stored frozen inside LDPE bottles until analysis. Possible contamination of the outer parts of the snow/ice cores, originating from metal impurities of the walls of the auger tube or during core handling (cutting, packing), was carefully considered. For this reason, a special decontamination procedure21 was applied in the cold clean laboratories strictly following the clean room procedure. This method enabled three annular concentric layers and one inner part core of the snow/ice sections to be analyzed separately to obtain radial concentration profiles. (15) Hong, S.; Candelone, J. P.; Patterson, C. C.; Boutron, C. F. Science 1996, 272, 246-249. (16) Boutron, C. F.; Go¨rlach, U.; Candelone, J.-P.; Bolshov, M. A.; Delmas, R. J. Nature 1991, 353, 153-156. (17) Barbante, C.; Turetta, C.; Capodaglio, G.; Scarponi, G. Int. J. Environ. Anal. Chem. 1997, 68, 457-477. (18) Van de Velde, K.; Boutron, C.; Ferrari, C.; Bellomi, T.; Barbante, C.; Rudnev, S.; Bolshov, M. Earth Planet. Sci. Lett. 1998, 164, 521-533. (19) Johnsen, J. S.; Dahl-Jensen, D.; Dansgaard, W.; Gundestrup, N. S. Tellus 1995, 47B, 624-629. (20) Vincent, C.; Vallon, M.; Pinglot, F.; Funk, M.; Reynaud, L. J. Glaciol. 1997, 43, 513-521. (21) Candelone, J.-P.; Hong, S.; Boutron, C. F. Anal. Chim. Acta 1994, 299, 9-16.
All the aliquots were analyzed directly after melting, except the two 7260- and 7760-year-old Greenland ice cores, which required a preconcentration by nonboiling evaporation.22 Using this procedure, the two samples were enriched by a factor of 52 and 77, respectively. Instrumentation and Measurement Parameters. An Element instrument (double-focusing ICPMS, Finnigan MAT, Bremen, Germany) was used. The instrument has predefined resolution settings of 300 (low-resolution mode, LRM), 3000 (mediumresolution mode, MRM), and 7500 (high-resolution mode, HRM) m/∆m (10% valley definition), which allow the signal of the analyte to be separated from those of most interferences. The LRM was used for the determination of Rh, Pd, and Pt, because at the expected concentration levels, from tens of femtograms per gram (1 fg g-1 ) 10-15 g g-1) for Rh to tens of picograms per gram for Pd, the highest ion transmission must be provided. The isotopes chosen (103Rh, 106Pd, 195Pt) were the most abundant and least interference-prone possible (see below). After melting inside the clean laboratory, the acidified aliquots were transferred in the sample introduction area of the instrument, where they were handled under a clean bench in order to minimize contamination from ambient air. Finally, samples were introduced into the plasma by a Micro Concentric Nebulizer (MCN-100, Cetac Technologies, Omaha, NB). The MCN-100 worked in the self-aspiration mode, to avoid possible contamination from Tygon peristaltic pump tubing and also fluctuations in the signal induced by the pump itself. The low sample uptake of the MCN-100 (40-80 µL min-1) was of paramount importance due to the low amount of sample material available (less than 1 mL in some cases). A homemade quartz double-pass spray chamber was cooled at 5.00 ( 0.01 °C with a thermostatic bath (RTE-300, Neslab Instrument B. V., Veldhoven, The Nederlands) or at 5 ( 1 °C with a different chiller (type 111 from Van der Heijden, Do¨rentrup, Germany). A quartz Fassel torch was used throughout the experiment. Sampling and skimmer cones, with an orifice diameter of 1.0 and 0.75 mm, respectively, were made of nickel. The instrumental conditions and measurement parameters are reported in Table 1. Thermally stabilized spray chambers are used to reduce the water content of aerosols entering the plasma and hence the consequent formation of polyatomic ions, which could interfere with analytes. To adopt the best conditions with respect to longterm stability of the signal, the performances of two different chillers were investigated by measuring the signal intensity of a 1 ng g-1 indium solution and the temperature of the spray chamber for a period of 140 min. The results, obtained using thermostatic baths in which the temperature was set at 5 °C but with different stabilities ((1 and (0.01 °C, respectively), are reported in Figure 1. It can be noted that temperature fluctuations in the spray chamber have a significant effect on the signal stability. In the case of the lower thermal stability ((1 °C, Figure 1a), variations in the ion signal are highly related to temperature fluctuations and are ∼(2.5% of the initial value (200 000 counts s-1). Moreover a reduction in the mean sensitivity of ∼4% of the initial signal is noted after 80 min, with a loss of the signal of ∼7000 counts s-1 h-1. After 80 (22) Go¨rlach, U.; Boutron, C. F. Anal. Chim. Acta 1990, 236, 391-398.
Table 1. Instrumental Conditions and Measurement Parameters for the Finnigan MAT Element forward power (W) gas flow rates plasma (L min-1) intermediate (L min-1) nebulizer
1300
14.5 0.7-0.9 optimized to obtain maximum signal intensity sample uptake rate (µL min-1) 40-80 washing time (min) 3 takeup time (min) 1 internal standards no ion sampling depth optimized to obtain maximum signal intensity ion lens settings optimized to obtain maximum signal intensity acquisition mode E-scan; electric scanning over small mass range 103Rh, 106Pd, 195Pt selected isotopes resolution required low (m/∆m ) 300) no. of scan 50 dwell time per acquisition point (ms) 10 no. of acquisition points per mass 50 segment (sample per peak) total acquisition time 0.5 s per mass segment and scan acquisition window (%) 100 search window (%) 100 integration window (%) 60
Figure 1. Long-term fluctuations of signal (s) and temperature (- - -) for a 1 ng g-1 indium solution using spray chamber cooling baths with different settings of thermal stability: (a) 5 ( 1 °C; (b) 5.00 ( 0.01 °C.
min, the cooling of the spray chamber was switched off and a temporary increase in the signal followed by a further, continuous loss in sensitivity was noted. In the second case (Figure 1b) where a temperature stability of (0.01 °C was used, no detectable fluctuations of the signal are registered over time. This long-term signal stability makes it Analytical Chemistry, Vol. 71, No. 19, October 1, 1999
4127
possible to avoid the use of internal standards and hence the possible problems of uncontrolled contamination. Considering that the thermal stability in the spray chamber can strongly affect the precision of the measurements, the cooling system with a more stable temperature control ((0.01 °C) was used with the real samples. The MCN gas flow rate was optimized daily using a 10 pg g-1 acidified (NIST HNO3, 1:200 diluted) ultrapure water solution of Co, In, and Pb in order to obtain the maximum signal response and good long-term stability.13 The maximum sensitivities at the optimum flow rates in the low resolution mode (m/∆m ) 300) were 180 000-320 000 counts s-1 per ng g-1 In, depending on the tuning conditions. When needed (usually weekly), accurate mass calibration was performed in LRM using a 1.0 ng g-1 multielement standard solution containing the following elements: Li, B, Na, Sc, Co, Y, In, Ba, Ce, Zn, Tl, and U. The stability of mass calibration was checked daily; drift in the masses of less than 0.02 amu was noted within 5 days. After a comparison between the standard addition method and the external calibration method, it was decided to use the external calibration curves for the routine quantification of the analytes, since the method provided faster and equally reliable results. Average values, intended as the sum of the intensities (counts s-1) divided by the number of acquisition points per mass segment (sample per peak) within the integration windows, are used. The intensity of the blank was subtracted from the intensities of the standard solutions and the signal increment plotted against the nominal concentrations of the standards. Linear calibration curves for 103Rh (y ) 185.3x + 0.26, r ) 0.994), 106Pd (y ) 121.1x + 0.60, r ) 0.985), and 195Pt (y ) 92.7x + 0.45, r ) 0.995), where y is the intensity (expressed as counts s-1) and x is the concentration of the element (in pg g-1), were then used for the quantification. Considering the time necessary for washing the system between two samples, the takeup time and the scanning time (see Table 1), the speed of analysis was about six samples per hour. Considering the good signal stability (see Figure 1b) and the very simple matrix (quite similar to ultrapure water), no internal standards were used, thus avoiding the possible introduction of contaminants into the samples. RESULTS AND DISCUSSION Selection of Isotopes and Interference Study. Although polar and alpine depositions such as snow, owing to their high purity, could be considered as ideal environmental matrixes for PGE determination, nevertheless they contain several other constituents that can potentially generate interfering species in the plasma conditions. Thus, the selection of the isotopes to be used for the analysis is a critical step in the setup of the methodology. Table 2 reports a list of the most abundant isotopes of Rh, Pd, and Pt, together with the most important interfering species, which can potentially form from snow and ice samples. It can be seen that several other constituents of the samples exert a key role in PGE determination since they can combine with other elements and interfere with the analytes. As regards selection of isotopes, 103Rh was chosen as the only available isotope for Rh, while 106Pd and 195Pt were preferred because they are the most abundant and at the same time offer 4128 Analytical Chemistry, Vol. 71, No. 19, October 1, 1999
the best expectation regarding the potential interferences. A specific study was carried out in order to ascertain whether the interfering species formed in the plasma could affect the determination of PGE in our samples. Moreover, it must be underlined that the method presented is feasible only for this ultrapure environmental matrix. As an example, in the analysis of soil samples, the quantification of Pd by using the 106Pd isotope is considerably hampered by the high Cd concentration in the same samples. 1. Interferences for Rh. The interferences having by far the most significant influence on Rh, due to the concentrations of the interfering species in snow and ice and their isotopic abundance, are the doubly charged 206Pb2+ and the polyatomic species 40Ar63Cu, 14N89Y, and 23Na40Ar40Ar for which resolutions of 1248, 8040, 30 000, and 11 300, respectively, are needed. Use of the highest resolutions achievable with the instrument (i.e., 3000 and 7500) to separate the 103Rh from the interference peaks was not considered because, as explained above, very low concentrations of Rh require the highest detection power of the low-resolution mode (m/∆m ) 300). To check all the possible interferences of 103Rh, five acidified (NIST HNO3, 1:100 diluted) ultrapure water solutions spiked with different concentrations (0, 10, 100, 500, and 1000 pg g-1) of each interference (Cu, Pb, Zn, Sr, Rb, Y, Na, Zr) were prepared and the relevant signals corresponding to the 103Rh peak were determined. The concentrations of trace elements in these synthetic solutions were chosen in order to cover the ranges expected in real samples.1,18 The results of this test (see Table 3) show that no signal increment, significantly different from the blank signal, was observed even for the highest interference concentrations, hence excluding any influence of all the possible interfering species on the analyte peak, at least up to 1000 pg g-1. 2. Interferences for Pd. As regards 106Pd, considering that of all potential interferences listed in Table 2 those that are more likely to give a contribution (referring to concentration in snow and isotopic abundances) are 106Cd, 40Ar66Zn, and 90Zr16O, the study of actual interferences was carried out in three steps. First a test similar to the one discussed above for Rh was carried out using multielement standard solutions containing all interferences except Zr. Then the contributions of Cd and Zr were studied independently. The results obtained with multielement standard solutions (0, 10, 100, 1000, and 5000 pg g-1 of Zn, Ge, Sr, Y, and Cd) are reported in Table 3 in terms of signal intensities (counts s-1). Data show a linear increment of the 106Pd signal with increasing interference concentration. The signal increment with respect to the blank signal (19 counts s-1) is also reported. To recognize the interference responsible for the signal increment, an independent experiment was carried out. Five Cd standard solutions (0, 10, 100, 1000, and 5000 pg g-1) were measured at mass 114. The Cd contribution to the signal of mass 106 was then calculated by the intensity of the 114Cd peaks (natural abundance 28.73%) and considering the natural abundance of the 106Cd (1.25%). The results of this experiment, shown in Table 3, demonstrate that the Cd contribution to the signal of mass 106 accounts for practically (within the experimental error) all the signal increment detected in the multistandard experiment,
Table 2. Potential Spectral Interferences That Could Affect the Determination of Ultratrace Levels of PGEs in Polar and Alpine Ice and Snow required analyte potential interference resolution isotope 103Rh
abundance (%) 100
species 38Ar65Cu 40Ar63Cu 36Ar67Zn 23Na40Ar40Ar 27Al36Ar40Ar 14N89Y 91Zr12C 87Sr16O 87Rb16O 85Rb18O 206Pb2+
104Pd
11.14
40Ar64Zn 38Ar66Zn 36Ar68Zn 40Ar64Ni 88Sr16O 87Sr17O 86Sr18O 87Rb17O 208Pb2+ 104Ru
105Pd
22.33
40Ar65Cu 36Ar69Ga 89Y16O 88Sr17O 87Sr18O 87Rb18O
106Pd
27.33
40Ar66Zn 38Ar68Zn 36Ar70Ge 90Zr16O 88Sr18O 89Y17O 106Cd
108Pd
26.46
38Ar70Ge 36Ar72Ge 40Ar68Zn 92Zr16O 91Zr17O 92Mo16O 108Cd
110Pd
11.72
40Ar70Ge 38Ar72Ge 36Ar74Ge 40Ar70Zn 36Ar74Se 94Mo16O 92Mo18O 92Zr18O 94Zr16O 93Nb17O 110Cd
194Pt
32.90
178Hf16O 177Hf17O 176Hf18O 176Yb18O 176Lu18O
195Pt
33.80
179Hf16O 178Hf17O 177Hf18O
196Pt
25.30
180Hf16O 180W16O 180Ta16O 178Hf18O 179Hf17O 196Hg
abundancea (%)
m/∆m
0.02 68.89 0.01 99.2 0.68 99.63 11.10 6.99 27.76 0.14 24.14 48.44 0.02 0.06 0.90 82.38 0.003 0.02 0.01 52.35 18.7 0.003 0.20 99.76 0.03 0.014 0.056 27.79 0.013 0.002 51.3 0.16 0.037 1.25 0.014 0.005 18.7 17.1 0.004 14.8 0.89 20.4 1.9 0.12 0.62 0.003 9.23 0.03 0.034 17.3 0.037 12.49 27.2 0.007 0.01 0.02 0.005 13.6 0.01 0.04 35.00 0.13 0.01 0.05 0.005 0.14
7200 8040 10100 11300 17240 30000 735000 102900 147000 17200 1248 8500 6200 9000 7900 33500 23600 21200 22100 1200 54700 7300 92000 27600 1000000 30900 28400 7200 6800 9300 26500 70600 58800 27900 8600 7300 6500 540000 216000 40000 1080000 6000 5300 7000 6300 7300 20000 366000 78500 26800 323251 52300 8100 9600 8800 9200 9200 8200 6900 8800 8348 8408 8692 8875 9803 226550
a For polyatomic species calculated as the product of the natural abundances of each isotope divided by 100; monatomic interferences, natural abundance.
revealing that no interference can be expected for Zn, Ge, Sr, and Y at least up to 5000 pg g-1. Consequently, to account for Cd interference in Pd determination, Cd was measured in all the samples analyzed (range 0.5-270 pg g-1) and its contribution to the mass 106 signal was subtracted before Pd quantification. It is to be remarked that for most of the samples the correction was insignificant with respect to the measurement error. To evaluate the contribution of possible 90Zr16O formation in the present working conditions (ZrO has a high oxide bond strength, 795 kJ mol-1, hence giving a large yield of MO+ in the plasma), eight acidified (1:100 diluted) ultrapure solutions were prepared with the following Zr concentrations: 0, 2, 10, 100, 200, 500, 800, and 1000 pg g-1. Monitoring the peaks at masses 90 and 106, an average MO+/M+ signal ratio of 0.0026 was calculated from the last four data (see Table 3), because under the experimental condition used throughout the work, the effect of oxide formation exceeded that of blank only at Zr concentration higher than ∼200 pg g-1. Considering also the natural abundances of 90Zr (51.45%) and the calculated abundance of 90Zr16O (51.32%) and taking into account the detection limit for Pd (0.08 pg g-1) and the different sensitivities for Pd (121.1 (counts s-1)/(pg g-1) and Zr (60 (counts s-1)/(pg g-1), calculated from Table 3, last four data), the minimum theoretical Zr concentration, detectable at mass 106 in the form of oxide, can be calculated as ∼62 pg g-1. This value roughlly agrees with the experimental limit observed from data reported in Table 3 (∼200 pg g-1). Considering the relatively high (with respect to PGEs) Zr concentration in bulk crustal material (237 µg g-1),23 a possible interference, in the form of refractory oxide (ZrO+) can be generated from the element naturally present in the snow and originating from a crustal source. The natural Zr background contribution from rock and soil dust in polar snow and ice samples can therefore be estimated from the Al concentration measured in the analyzed samples (median value, 20.5 ng g-1) and the Zr/ Al mass ratio in bulk crustal material (3.1 × 10-3).23 Combining these values, a natural background concentration of ∼63 pg g-1 of zirconium was calculated. However, Zr can also be found in the emission from modern catalytic converters,24 since it is added (in a ratio of 10:1 with respect to Pd 25) to the catalytic support to improve thermal stability and poison resistance. Considering that the effect of ZrO+ formation influences the peak at mass 106 only at a Zr concentration above ∼200 pg/g, (see Table 3), then the resulting contribution to oxide formation from this latter source is negligible in most cases. 90Zr was nevertheless monitored in all samples (range, 0.3-190 pg g-1) and in the case of relatively high Zr concentration (e.g., in the Alpine samples or old Greenland ice, where the soil and dust contribution is sometimes high) a correction, in accordance with the above measured ZrO+/Zr+ signal ratio, was carried out. Moreover, the interference of ZrO+ can be generated also from the element naturally present in the snow and originating from the cosmic dust, which contribute to the total PGE inventory in the snow (see below). In this case, considering that the Zr concentration in the cosmic dust is only 7 times higher than the Pd concentration (3940 and 560 ng g-1 for Zr and Pd, respectively),26 then the possible formation of ZrO+ from the cosmic source is negligible. (23) Wedepohl, K. H. Geochim. Cosmochim. Acta 1995, 59, 1217-1232. (24) Helmers, E. Chemosphere 1996, 33, 405-419. (25) Wayne, D. M. J. Anal. At. Spectrom. 1997, 12, 1195-1202.
Analytical Chemistry, Vol. 71, No. 19, October 1, 1999
4129
Table 3. Intensity Values (Counts s-1) of a Blank Solution Spiked with Different Amounts of Potential Interfering Elementsa concentration of the interferences (pg g-1) isotope interferences 103Rh
Cu, Pb, Zn, Sr, Rb, Na, Zr, Y
106Pd
Zn, Ge, Sr, Y, Cd Cd 90Zr 90Zr16O
195Pt
179Hf 179Hf16O
a
0
signal
2
10
100
200
500
800
1000
5000
60 (10)a
51 (9)
47 (6)
19 (7)
36 (3)
144 (18)
1166 (49)
5830 (137)
17 352 15 440 (15) 22 (3)
125 2959 129 4846 (575) 10924 (533) 30130 (765) 47362 (393) 23 (2) 27 (3) 81 (9) 128 (2)
1147 26381 1148 59963 (598) 151 (12)
5811 134096 5834
signal increment signal at mass 114 24 calculated 106Cd 15 (0.3) 91 (4) 23 (2) 19 (2) 3 (2) 5 (6)
54 (6)
46 (8)
144 (18) 1984 (48) 7726 (642) 16250 (373) 44278 (236) 66922 (690) 86066 (1274) 6 (2) 8 (3) 11 (1) 16 (3) 38 (2) 53 (13) 78 (4)
Detailed explanation in the text. b In parentheses, SD (n ) 3).
3. Interferences for Pt. The most important interference that could affect the determination of platinum in the chosen isotope (195Pt) arises from 179Hf 16O. Hf is present in the continental crust at a concentration of ∼14 000 times higher than Pt 23 and hence the contribution of oxide formation could be relevant. In this case too, as for Pd, the high oxide bond strength (801 kJ mol-1) can lead to the formation of a considerable amount of MO+. The MO+/ M+ signal ratio, calculated as before (see Table 3), is 0.0009. From the experimental data of Table 3, it can be noted that in this case the effect of oxide formation exceeded the signal of the blank at a Hf concentration above ∼10 pg g-1. Considering also the natural abundance of 179Hf (13.63%) and the calculated abundance of 179Hf 16O (13.60%) and taking into account the detection limit for Pt (0.008 pg g-1) and the different sensitivities for 195Pt (92.7 (counts s-1)/(pg g-1) and for 179Hf (85 (counts s-1)/(pg g-1), calculated from Table 3, last four data), the minimum theoretical Hf concentration, detectable at mass 179 in the form of oxide, can be calculated as ∼10 pg g-1. This result is in substantial agreement with the experimental value observed in Table 3 (∼10 pg g-1). The natural contribution to Hf concentration can include different sources such as rock and soil dust, volcanic emissions, cosmic dust fallouts, and biomass burning. A complete and satisfactory estimation of these sources cannot be carried out because of the lack of data on the natural emission of PGE. The only natural contribution that can be roughly estimated is that coming from rock and soil dust, from the Al (median value, 20.5 ng g-1) and the Hf/Al mass ratio in bulk crustal material (7.5 × 10-5).23 Combining these values, a natural background crustal concentration of ∼1.5 pg g-1 hafnium was calculated. Considering that the effect of oxide formation influences the peak at mass 195 when the concentration of Hf is ∼10 pg g-1 (see Table 3), then the resulting contribution to oxide formation from this latter source is negligible in most cases. Nevertheless, Hf concentration was measured in all samples (range 0.01-2.1 pg g-1) and whenever necessary a correction to the Pt concentration was carried out. In particular, in the case of the two older samples (dated to 7260 and 7760 years BP), Pt concentrations of 0.008
and 0.015 pg g-1 were found, with negligible Hf contribution. In fact, Hf concentrations in the analyzed enriched samples were 0.62 and 1.34 pg g-1, respectively. As for Pd, the interference of the oxide (HfO+) can originate also from the cosmic dust. In this case, considering that the Hf concentration in the cosmic dust is ∼10 times lower than the Pt concentration (104 and 990 ng g-1 for Hf and Pt, respectively),26 then the possible formation of HfO+ from the cosmic source is negligible. Blanks, Detection Limits, and Repeatability. Considering the purity of the samples to be analyzed, particular care was taken in the evaluation of the blanks and in the choice of reagents. A potentially important contribution to the detected PGE concentrations could originate from acidification of samples and standards. This contribution was evaluated by measuring seven ultrapure Milli-Q water solutions with increasing contents of HNO3 NIST (ultrapure water, 1:1000, 1:500, 1:200, 1:50, 1:20, and 1:10 diluted, respectively). Three independent measurements were carried out in each sample. No signal intensity variations (changes always below the standard deviations) were observed, highlighting that any possible contribution from acidification is negligible. The PGE concentrations in the ultrapure Milli-Q water, determined from the six-point calibration curve, and the standard deviations from n ) 30 (in brackets) were (in pg g-1) as follows: Rh 0.030 (0.007), Pd 0.31 (0.026), and Pt 0.044 (0.0027). Detection limits (DL) were calculated as 3 times the standard deviation of the ultrapure water response (n ) 30), in accordance with IUPAC recommendations.27 They were 0.02, 0.08, and 0.008 pg g-1, for Rh, Pd, and Pt, respectively. An estimation of the instrumental repeatability of the data obtained from 30 consecutive measurements of a real Alpine snow sample (Colle Gnifetti core, sample 9.1, depth 6.0 m) gave the following mean values (in pg g-1) and the relative standard deviation (in parentheses): Rh, 0.06 (27%); Pd, 0.9 (28%); Pt, 0.2 (29%). These relatively large deviations are due to the extraordinarily low concentrations measured. Accuracy. The accuracy of trace element determinations at the low and subpicogram per gram level is very difficult to estimate owing to the different relative contribution of the various error
(26) Anders, E.; Grevesse, N. Geochim. Cosmochim. Acta 1989, 53, 197-214.
(27) Long, G. L.; Winefordner, J. D. Anal. Chem. 1983, 55, 712A-724A.
4130 Analytical Chemistry, Vol. 71, No. 19, October 1, 1999
Table 4. Recovery Data (pg g-1) for the Determination of Rh, Pd, and Pt in the 86.4-m-Deep Do ˆ me de Gou ˆ ter, Mont Blanc Sample
isotope 103Rh 106Pd 195Pt a
sample conc
added
0.29 (0.06)a 6.7 (0.7) 0.64 (0.06)
2.0 2.0 2.0
spike A found 1.7 (0.3) 1.6 (0.6) 1.7 (0.3)
added 5.0 5.0 5.0
spike B found 3.9 (0.9) 4.8 (0.4) 4.4 (0.6)
added 10 10 10
spike C found 8.7 (1.0) 10.0 (1.0) 9.4 (0.9)
added 20 20 20
spike D found 17.6 (1.5) 21.0 (1.0) 18.4 (1.1)
In parentheses, SD (n ) 3).
Figure 2. PGE radial concentration profiles for an Alpine snow core (Colle Gnifetti, length 30 cm; diameter, 10.5 cm; depth, 9.6 m). (a) Rh; (b) Pd; (c) Pt. Error bars as SD are also reported.
sources in the entire analytical procedure. Among these, the accurate preparation of very diluted synthetic multielement standards used for the calibration curves poses the most severe problems. Due to these uncertainties, considering that a certified reference material for the snow and ice is presently not available, we tried to estimate the accuracy of the measurements by spiking a real sample with known amounts of Rh, Pd, and Pt and determining the recoveries. Table 4 reports the recovery data as obtained for the 86.4-mdeep Doˆme de Gouˆter (Mont Blanc) sample. This sample was spiked with 2 (spike A), 5 (spike B), 10 (spike C), and 20 (spike D) pg g-1 Rh, Pd, and Pt. All the recoveries can be considered in agreement with the added concentration values (deviations are generally within the experimental error, RSD ∼30%) even if they are generally slightly lower than the expected one (see Rh in particular). Radial Concentration Profiles. Analysis of concentric layers of snow/ice cores obtained from the decontamination procedure21 enabled us to obtain information on how far from the outside the outer contamination has penetrated and whether the inner parts of the core are free of contamination or not. A concentration plateau, in at least two consecutive layers in the central part of the core, indicates that these inner parts are free from the outside contamination; the plateau value then represents the original concentration in the snow or ice core. If, on the contrary, a continuous decrease of concentration is observed toward the center, it testifies that outside contamination has penetrated in the center of the core. The concentration of the inner part will then represent an upper limit of the real concentration in the original snow and ice.
As far as we know, this study is the first ever carried out for Rh, Pd, and Pt. As an example, Figure 2 shows the radial concentration profiles obtained for a Colle Gnifetti snow core collected at a depth of 9.6 m (core section was 10.5 cm in diameter and 30 cm in length). These profiles evidence that, contrary to results obtained for other trace elements,21 contamination problems here are not present. Differences in concentration along the outside-inside profile are probably due to the uncertainty in the measurements, as evidenced by the reported error bars. PGE Concentrations in Polar and Alpine Samples. Summary statistics of PGE concentrations in Greenland, Antarctic, and Alpine snow and ice samples covering different time periods are reported in Table 5. In the case of snow/ice cores, the data refer to the decontaminated innermost part of them. The wide variability of concentration values is related to seasonal effects, which strongly affect the heavy metal concentrations in polar and alpine snow.13,18 In a longer time scale, comparing the PGE concentrations of very old Greenland ice dating back seven millennia (Holocene), with the values of recent samples as detected in Greenland (consider, e.g., median values), it can be seen that the concentrations of these elements are several times higher than the background natural Holocene levels. It is to be noted that the extremely low PGE concentrations for Greenland ice core dating from seven millennia ago (see Table 5) must be totally imputable to natural sources. By combining the aluminum concentrations measured in old Greenland ice samples (4 ng g-1)28 with the PGE/Al ratios in bulk crustal (28) Hong, S.; Candelone, J.-P.; Patterson, C. C.; Boutron, C. F. Science 1994, 265, 1841-1843.
Analytical Chemistry, Vol. 71, No. 19, October 1, 1999
4131
Table 5. Summary Statistics for Concentrations of PGEs in Greenland, Antarctic, and Alpine Snow and Ice Samples Rh concn (pg g-1)
Greenland ice core samplesa Greenland recent snow core samples Greenland recent snow pit samples Antarctic surface snow samples Dome de Gouˆter ice and snow samples Colle Gnifetti snow samples a
period of deposition
no. of samples
7260; 7760 BP
2
median (min - max)
mean (SD)
Pd concn (pg g-1) median (min - max)
mean (SD)
Pt concn (pg g-1) median (min - max)
mean (SD)
1969-1988
22
7 × 10-4 ((5-9) 7 × 10-4 0.01 (0.01-0.01) 0.01 (0.00) 0.01 (0.008-0.015) 0.01 (0.055) × 10-4) (3 × 10-4) 0.04 (0.02-0.10) 0.04 (0.02) 0.38 (
