Geochemical Journal, Vol. 20, pp. 233 to 239, 1986

Determinations of osmium isotope ratios in iron and iridosmines by ICP-MS

AKIMASA MASUDA, TAKAFUMI HIRATA and HIROSHI SHIMIZU

Department of Chemistry, Faculty of Science, The University of Tokyo, Hongo, Tokyo 113, Japan

(Received August 15, 1986: Accepted September 10, 1986)

Osmium isotopes in three Os reagents, four terrestiral Os-Ir alloys (iridosmine) and two iron mete orites have been measured on an inductively coupled plasma mass spectrometer (ICP-MS), with the preci sion of around 1% for 1870S/1920S. Comparisons are made of our isotopic ratios with the previous ones. Apart from the variation of 1870S abundance as naturally expected, there seems to be a slight systematic deviation for "'Os. Also some discussions are given to 184Os. The Os isotopic ratios for three iridosmine samples from Hokkaido, Japan, have been found quite similar to each other. The values for the Hokkaido iridosmines are 3-4% higher relative to the straight "'Os growth line obtained by Allegre and Luck (1980) . Finally, the isotopic "standard" abundances (%) for Os are presented by us: 1840s, 0.019; 1860s, 1.589; 187Os, 1.519; 1880s, 13.24; 1890s, 16.37; 190Os , 26.30; 192Os, 40.96. These give us the Os atomic weight = 190.24.

INTRODUCTION determined the 117Re half-life ((4.35 ± 0.13) X 1010y) with these measurements. The half life Herr et al. (1961) started Re-Os cosmo of 117Re is similar to that of 8'Rb (4.88 X 101°y). chronology on the basis of the g decay of Further, Os isotope measurements by accelera 187Re, but the difficulty in Os isotope measure for mass spectrometer have been reported on ments prevented the popular use of this method. meteoritic and terrestrial samples (Teng et al., Recent developments in a variety of mass spec 1986). trometry have allowed us easier access to Os Re and Os are siderophile or chalcophile isotope measurements. Luck et al. (1980) and elements. Further, it has been pointed out that Allegre and Luck (1980) established Os isotope Re/Os ratio has a large fractionation factor measurements with secondary ion mass spectro during continental crust/mantle fractionation. meter (SIMS) and reported Re-Os age for These geochemical features indicate the im meteorites. Since these works, geochemical and portance of Re-Os systematics as an isotope cosmochemical studies on Re-Os systematics tracer and geo and cosmo-chronometer. Thus, have proceeded on the basis of Os isotope data it can be said that the recent developments in obtained by SIMS (Luck and Allegre, 1982, mass spectrometry are strongly promoting the 1983, 1984; Luck and Turekian, 1983; Palmer further progress of the Re-Os method. and Turekian, 1986). Luck and Allegre (1983) In this paper, we will report Os isotope reported the half life value of (4.56 ± 0.12) X ratios in iron meteorites and iridosmines as well 1010y based on the meteoritic isochron. On the as Os reagents measured by ICP-MS. ICP-MS other hand, Lindner et al. (1986) measured Os uses inductively coupled plasma (ICP) as an isotope ratios in HReO4 by two different types atmospheric ion source for a quadrupole mass of mass spectrometers, an inductively coupled spectrometer (Gray, 1975, 1985; Gray and Date, plasma mass spectrometer (ICP-MS) and a laser 1983; Houk et al., 1980). The most prelimina microprobe mass analyzer (LAMMA). They ry work of ours was presented at the 1985

233 234 A. Masuda et al.

Annual Meeting of the Geochemical Society of affect Os' mass spectrometry. Japan. ICP-MS used in this study was VG Isotopes PlasmaQuad. Sample solution of 1 % HNO3 was introduced to ICP-MS through the nebuli SAMPLES AND EXPERIMENTAL METHOD zer. All of Os isotopes were measured; 184, Osmium tetroxide (Os04) reagents prepared 186, 187, 188, 189, 190 and 192. 192Oswas by Strem Chemicals Inc., E. Merck Co. Inc. and used as a reference isotope in this study. Os Nakarai Chemicals Ltd. were used for our exami isotopic abundance obtained by Nier (1937) nation. Os isotopes in these reagents were was used for correction of mass discrimination, measured for 1 ppm Os solution (1 % HNO3 solu which is currently inevitable as shown later. tion). Besides, Os isotopes were measured on The mass discrimination as observed in ICP-MS two iron meteorites, Canyon Diablo and Tlaco rests mostly on the temperature of RF/DC tepec. The iron meteorites were decomposed generator of quadrupole mass spectrometer and by aqua regia (1:3 mixture of HNO3 and HC1) on the condition of electrode bias supplied to at 80-100°C. In the decomposition process, the ion lens system. Generally, for the quadru Os was oxidized to volatile Os04. The evapo pole mass spectrometer like this, the mass dis rated Os04 was trapped with distilled water. crimination effect is inevitable and is not a The trapped OsO4 solution was free from Pt, W strictly linear function with the mass number. and Re. Four iridosmine samples were also However, for the rather short mass range dealt studied for Os isotope ratios; three samples are with here and within the precision under con from Hokkaido, Japan and another is from the sideration, one can assume the mass discrimina Urals, USSR. Iridosmine is a terrestrial Os-Ir tion effect as a linear function with the mass alloy. Iridosmine samples were decomposed by number. sodium peroxide (Na202) in a zirconium cru cible. After decomposition, Os concentration of RESULTS AND DISCUSSION each sample solution was adjusted to about 1 ppm. We have confirmed that (Zr2J ions do Os reagent not appear, but very weak Zr+ ions are recog Figure 1 shows two examples of Os isotopic nized in such a product. If (Zr2)+ appear, they measurement for Strem Os reagent. Ordinate in

STREM CHEM. lppm Os

c 1.05 0

0 a c E.c U) 1 0 U) U) 0

0.95

184 186 187 188 189 190 192 Mass number

Fig. 1. Mass discrimination factor relative to Os isotope ratios of Nier (1937), employing 192 Os as a " reference l1 nuclide. Determination of Os isotope ratios 235

Fig. 1 refers to ratios of mOs/192Os (m = 184, and Table 2 shows Os isotopic abundances. The 186, 187, 188, 190 and 192) divided by the Os isotopic data by Nier (1937) and Luck and corresponding Os isotope ratios obtained by Allegre (1983) are also presented in Tables 1 Nier (1937). Three points of "'Os/'9'0s, 188Os/ and 2. Figure 3 shows Os isotope data for Os 192Os and 190Os/1920s fall close to a straight reagents , corrected by mass discrimination line and a point of "90s/"'Os appears to deviate factor as defined by the least-square line for from this line to a certain extent. This can be 186Os/1920s, 188Os/192Os and 190Os/1920s. As understood as implying that 1110s isotopic shown in Fig. 3 and Table 1, our data for abundance obtained here differs from that by 186Os/1920S, 188Os/1920s and 190Os/1920s agree Nier. well with those obtained by Nier (1937) . and Correction factors of mass discrimination Luck and Allegre (1983). As for 189Os/192Os for each isotope can be obtained from the ratio, our value is close to that of Luck and straight line drawn by a least-squares method Allegre, but it is 1.7% higher than that of with respect to points of 1860S/1920S, 188Os/ Nier. Precision of 1110S/192Os ratio is worse 192Os and 190Os/192Osas shown in Fig. 1. When than those for other Os isotope ratios, because dividing the actually observed ratios by the of very low isotopic abundance of 184Os. Any correction factors mathematically given by the how, 1140S/192Os ratio obtained here is close to least-square line, one can obtain the corrected the value by Nier (1937), but 184Os/192Os value ratios. Figure 2 shows "'Os/"'Os and 1860s/ obtained by Luck and Allegre (1983) is about "'Os ratios thus corrected , for Strem Os04 20-40% higher than that obtained by Nier and reagent during this work. In this diagram, the in this study. It is noted here that 184Wis the precision of 1870s/1920s is about 0.4% (2am). major isotope (30.7%) in tungsten and 184Os From Fig. 2, it can be seen that the precision (0.018%) is subject to the interference by W. for 187Os/1920s is much worse than that for The data cited in Table 1, from the work by 11110S/192Os. This is due to a memory effect, Luck and Allegre (1983) is for their Merck because, while measuring the Strem Os reagent, standard. we measured intermittently other two reagents Specifically, in our measurements, any W' having different 1170S/192Os ratios from that peaks have not been observed for Os reagents. of Strem. If more caution is taken against However, when using the solutions simply such an effect, better data will be obtained. prepared by dissolving the iridosmines from Table 1 lists Os isotopic ratios against 1920s Nuppu River and Nizhni Tagil, we could recog 1860s/1920a=0.03881±0.00002

0.03885 ~TTT 0.03880

0.03875

1870S/1920S.0 .04862±0.00022

0.04910

I 1 i I I T 0.04860 I I T 1 1 l~ I I 0.04810 L 1870s/192Os and 1860s/!92 Fig. 2. 1 Os ratios in repeated measurements of Strem Os reagent. Errors are 2am. 236 A. Masuda et al.

Tabble 1. Os isotope ratios against 1920s for Os reagents, together with isotopic ratios by Nier (1937) and Luck and Allegre (1983)

1840s 1860s 1870s 1880s 1890s 19005

Strem 0.00042 0.03881 0.04862 0.3238 0.4002 0.6428 ±0.00002 ±0.00002 ±0.00022 ±0.0005 ±0.0005 ±0.0017 Nakarai 0.00050 0.0388 0.0463 0.3223 0.4002 0.6421 ±0.00010 ±0.0014 ±0.0010 ±0.0016 ±0.0028 ±0.0020

Merck 0.00045 0.0387 0.0371 0.3237 0.3992 0.6416 ±0.00008 ±0.0002 ±0.0010 ±0.0012 ±0.0038 ±0.0016

Average* 0.00046 0.0388 0.3233 0.3998 0.6422 (I) ±0.00004 ±0.0001 ±0.0009 ±0.0006 ±0.0007 Average" 0.00042 0.0388 0.3237 0.4000 0.6421 (II) ±0.00002 ±0.0001 ±0.0006 ±0.0002 ±0.0007

Nier 0.00049 0.0388 0.0393 0.3244 0.3927 0.6439 Luck & 0.00059 0.03904 0.03687 0.32439*** 0.39679 0.64382 Allegre ±0.00006 ±0.00002 ±0.00002 ±0.00011 ±0.00026 Errors are 2 0m. * Average (I): arithmetical mean. "Average (II): weighted mean, taking into account the statistical fluctuations in measure ment. *** Luck and Allegre (1983) used 1920x/18805= 3.08271 for normalization. nize the W+ peaks. Their peak heights were some uneasiness may remain in discussing "'Os such as to affect to some extent only "'Os. In relative to 186Os. At present, however, discus the case a reliable value for 114Os relative abun sions about "'Os will be given in terms of "'Os/ 186Os dance is needed for iridosmine, we are currently , following the way currently adopted. employing the separation of Os vapor from the Isobaric effect on "'Os (1.59%) by 186W simply dissolved sample solution. For mete (28.4%) could not be so serious as the similar orites, we also carried out the Os vapor separa one on 1140s by 184Wso long as the ordinary tion. caution is taken in separation of Os from W. It might merit notice that 190Pt decays into The W-Os interference can be detected by "'Os with the half -life of 5.9 X 10"y. The sweeping the mass range involving tungsten isotopic abundances of "'Pt (0.0127%) and isotopes. But the effect as noted here might "'Os (1 .59%) and the decay constant con become an additional reason to let us prefer cerned suggest that the possible isotopic varia 187Os/188Osratio to "'Os/"'Os in geochemical tion of the latter nuclide should be far smaller discussions. than the current detection limits but can be detected in unusually Pt-rich minerals. Hence

Table 2. Osmium isotopic abundances (%)

184s 186s 1870s 188s 1890s 1900s 1925 This study* A 0.019 1.582 1.982 13.18 16.30 26.18 40.76 B 0.019 1.589 1.519 13.24 16.37 26.30 40.96 Nier * * 0.018 1.59 1.64 13.2 16.1 26.4 41.0 Luck & Allegre*** 0.024 1.599 1.510 13.29 16.25 26.37 40.96 * These isotopic abundances were calculatedfrom the arithmeticalmean of the Os ratios in Table 1, except for 1870s/1920s;"A" was based on the 1870s/1920sratio for Strem Os reagent and "B"on the 1170S/192Os ratio for MerckOs reagent. **Nier (1937). *** Luck and Allegre (1983) for Merck Os reagent. Determination of Os isotope ratios 237

Iron meteorites and iridosmines Table 3. Osmium isotopic ratio for iron meteorites and Os isotopic ratios for iron meteorites and iridosmines iridosmines are shown in Table 3. 1870s/1860s 1860S/1920S 18705/19206 18705/18606 ratios are 1.066 and 1.033 for Canyon Diablo Iron * and Tlacotepec, respectively. Luck and Allegre Canyon Diablo 0.0391 0.0417 1.066 ±0.0012 ±0.0008 ±0.038 (1983) reported 1110S/116Os ratios of 1.115 and Tlacotepec 0.0389 0.0402 1.033 1.113 for Canyon Diablo and of 0.980, 0.994 ±0.0006 ±0.0008 ±0.020 and 1.008 for Tlacotepec. We cannot rule out Iridosmine * Kodaira, Hokkaido 0.0387 0.0415 1.072 a possibility that the numerical differences ±0.0002 ±0.0006 ±0.016 between those by Luck and Allegre and ours Nuppu River, Hokkaido 0.0388 0.0414 1.067 may be due to the heterogeneity in Re/Os ±0.0010 ±0.0006 ±0.032 Usotannai, Hokkaido 0.0387 0.0412 1.065 ratios in the specimens. ±0.0004 ±0.0004 ±0.016 Three iridosmine samples from Hokkaido Nizhni Tagil, The Urals 0.0388 0.0393 1.013 ±0.0002 ±0.0005 ±0.006 (Kodaira, Nuppu River and Usotannai) are considered to be associated with ultra-mafic Errors are 2 am. rocks in the Kamuikotan metamorphic belt. * Amounts of samples used are ca. I g for iron meteorite and ca. 1 mg for iridosmine. Metamorphic age for the Kamuikotan belt is thought to be about 100 m.y. Iridosmine from the Urals (Nizhni Tagil) is associated with ultra reported "'Os/"'Os ratio for "osmiridium" mafic rocks (300-400m.y.) in the "Hercynian samples of various ages from 90 to 2,700m.y. ophiolite belt". Allegre and Luck (1980) Their conclusion is that the 1170S/"'Os variation with time can be fitted by a straight line cor responding to 187Re/186Os= 3.15 for the mantle,

1.4 This work 0 STREM CHEM. comparable to Cl carbonaceous .

• NAKARAI, Our 1870s/186Osratio (1.0 13) for the sample from the Urals is somewhat lower than that 0 MERCK, (1.036 ± 0.004) reported by Allegre and Luck 1.3 • Luck and Allegre(1983) (1980). Our value appears to fall on the straight m 0 line mentioned above, while their value deviates 0 u 0 somewhat from it. 1.2 ro w As seen in Table 3, the 11170S/186Os ratios m (1.065-1.072) for three Hokkaido iridosmines wd are quite similar to each other. Assuming the z a 1.1 fixation of 1870S/116Os ratio 100m.y. ago cor

v d responding to the metamorphic age, we consider N M that the Hokkaido iridosmine ratio is 3-4% e w 0 higher as compared with the Allegre-Luck's z 1 j straight line. If the fixation age is older than the metamorphic age, the difference from the straight line would tend to increase. This dif

0.9 ference could be explained either as reflecting

184 186 187 188 189 190 192 some possible variation of 187Os/186Osratio in

Mass number the mantle source, or as due to a small extent of contamination by the crustal source. The Fig. 3. Relative Os isotope ratios to 1920s for Os explanation in terms of the latter effect might reagents, together with Os isotope ratios obtained by Luck and Allegre (1983). Errors are tam. (For correc be unfavorable in view of the considerably tions for mass discrimination, see text.) high homogeneity for three samples studied 238 A. Masuda et al.

here, but it would be premature to give con work was supported by granti-in-aid for Scientific clusive words to this matter yet. One cannot Research from the Ministry of Education, Science and rule out a third possibility such as double Culture of Japan. stage evolution of the mantle and even a fourth one related with the systematic aberration in REFERENCES measurements. But, in view of the variation in Allegre, C. J. and Luck, J.-M. (1980) Osmium iso meteoritic 187Os/186Osratios as observed by topes as petrogenetic and geological tracers. Earth Luck and Allegre (1983) and in the present Planet. Sci. Lett. 48, 148-154. study for Canyon Diablo and Tlacotepec, it is Gray, A. L. (1975) Mass-spectrometric analysis of most likely that the deviation observed here is solutions using an atmospheric pressure ion source. due to some heterogeneity of Re-Os ratio in the Analyst 100, 289-299. mantle source. Gray, A. L. (1985) The ICP as an ion source origins, achievements and prospects. Spectrochimica Acta Meanwhile, Morgan (1985) discussed the 40B, 1525-1537. Os/Re ratios for the mantle source, in relation Gray, A. L. and Date, A. R. (1983) Inductively with the Os/Re ratios in classified coupled plasma source mass spectrometry using con groupings. He maintains that the appropriate tinuum flow ion extraction. Analyst 108, 1033 Os/Re value for the earth's mantle source 1050. should be 12.06 by weight, corresponding to Herr, W., Hoffineister, W., Hirt, B., Geiss, J. and Houter 187Re/186Os= 3 mans, F. G. (1961) Versuch zur Datiering von .32. If similar calculation is Eisen Meteoriten nach der Rhenium-Osmium made for the Hokkaido iridosmine, we can get Methode. Z. Naturforsch. 16,1053-1056. 187Re/186Os= 3.27, which falls between 3.15 Houk, R. S., Fassel, V. A., Flesch, G. D., Svec, H. J., (Allegre and Luck, 1980) and 3.32 (Morgan, Gray, A. L. and Taylor C. E. (1980) Inductively 1985). coupled argon plasma as an ion source for mass In summary, isotope ratios of osmium for spectrometric determination of trace elements. Anal. Chem. 52,2283-2289. osmium tetroxide reagents, iridosmines and Lindner, M., Leich, D. A., Borg, R. J., Russ, G. P., iron meteorites have been measured by ICP-MS, Bazen, J. M., Simons, D. S. and Date, A. R. (1986) with the relative precision of around 1 % for Directly laboratory determination of the 187Re half 187Os/192Osratio . Some minor differences have life. Nature 320, 246-248. been observed in Os isotopic abundances among Luck, J.-M. and Allegre, C. J. (1982) The study of the data by Nier (1937), Luck and Allegre molybdenites through the 187Re-1870s chronometer. Earth Planet. Sci. Lett. 61, 291-296. (1983) and this study; slightly lower abundance Luck, J.-M. and Allegre, C. J. (1983) 187Re-1870s of 1890s in the data of Nier and higher abun systematics in meteorites and cosmochemical con dance of "'Os in the data of Luck and Allegre. sequences. Nature 302, 130-132. We think that the isotopic abundances indicated Luck, J.-M. and Allegre, C. J. (1984) 187Re-18 70s by B in Table 2 are recommendable as standard investigation in sulfide from Cape Smith komatiite. values. Based on this data, one can obtain the Earth Planet. Sci. Lett. 68, 205-208. Luck, J.-M. and Turekian, K. K. (1983) Osmium-187/ atomic weight for Os to be 190.24. osmium-186 in manganese nodules and the Creta ceous-Tertiary boundary. Science 222,613-615. Acknowledgements-We would like to express our Luck, J.-M., Brick, J.-L. and Allegre, C. J. (1980) gratitude to Dr. A. Kato, National Museumof Natural 187Re-1870s systematics in meteorites: early chro History, Japan, for his helpful information. Iridosmine nology of the solar system and age of the Galaxy. from Nizhni Tagil, the Urals, USSR, was offered by Nature 283, 256-259. Dr. P. T. Dunn, National Museum of Natural History, Morgan, J. W. (1985) Osmium isotope constraints on Smithsonian Institution, WashingtonD. C. The kind Earth's late accretionary history. Nature 317, 703 arrangements in offering the Tlacotepec meteorite 705. specimensfrom Dr. E. J. Olsen, Field Museumof Nat Nier, A. 0. (1937) The isotopic constitution of ural History and Dr. C. O'Neil, American Museum of osmium. Physical Review 52,885. Natural History, are gratefully acknowledged. This Palmer, M. R. and Turekian, K. K. (1986) 1870s/ Determination of Os isotope ratios 239

186O5 in marine manganese nodules and the con H. E. (1986) Osmium isotope ratios determined in straints on the crustal geochemistries of rhenium and low level samples using accelerator mass spectrometry osmium. Nature 319, 216-220. and a nickel sulphide preconcentration method Teng, R., Fehn, U., Elmore, D., Kubik, P. W. and Gove, (Abstract). Terra cognita 6, 122.