Osmium Isotope Evidence for an s-Process Carrier in Primitive Chondrites

See allHide authors and affiliations

Science  19 Aug 2005:
Vol. 309, Issue 5738, pp. 1233-1236
DOI: 10.1126/science.1115053


Osmium extracted from unequilibrated bulk chondrites has isotope anomalies consistent with an insoluble s-process carrier, termed Os(i) here. Osmium from metamorphosed bulk chondrites does not have isotope anomalies, implying that the Os(i) carrier was destroyed by metamorphism. The isotopic homogeneity of metamorphosed bulk chondrites is consistent with extremely effective mixing of presolar grains from varied sources in the nebula. Osmium in the Os(i) carrier is likely from nucleosynthetic sites with a neutron density about two to four times as high as that of the average solar s-process Os.

Presolar grains (such as SiC and nanodiamonds) are prevalent in unequilibrated chondrites and preserve material from nucleosynthetic processes occurring in the stars from which these grains formed (1, 2). The degree to which these grains and solar material were mixed in the solar nebula is uncertain but important for discerning solar system processes. Isotopic anomalies have been identified in presolar grains and Ca-Al–rich inclusions (CAIs) in chondrites (14). We measured Os isotopes to assess heterogeneities in bulk chondrite meteorites: 184Os is produced by the p process only; 186Os [p-process contribution 1.1% relative (5)] and 187Os are produced by the s process and radiogenic decay from 190Pt and 187Re, respectively; 188Os, 189Os, 190Os, and 192Os are produced partly by the s process but are dominanted by the r-process Pt peak (68) (Fig. 1). Stellar nucleosynthetic processes release their products to the interstellar medium as both gas and grains. Osmium condenses as an ultrarefractory metal alloy with Re and W, and behaves as one of the most refractory elements during condensation of stellar outflows. Ultrarefractory grains are more likely to survive nebular thermal processing. We studied seven ordinary chondrites: Chainpur LL3.4, Parnellee LL3.6, Sharps H3.4, St. Marguerite H4, Forest Vale H4, Allegan H5, and Weston H5; three enstatite chondrites: Daniels Kuil EL6, Yilmia EL6, and Indarch EH4; and three carbonaceous chondrites: Tagish Lake C2, Ornans CO3, and Allende CV3. Together, these chondrites span a wide range of compositional groups and metamorphic grades.

Fig. 1.

The s-process pathway in the W-Os region. Stable isotopes are indicated in black letters with their isotopic abundances given below the nuclide symbol. Long-lived radioactive elements are indicated in white letters on gray background, and short-lived radioactive elements are indicated in gray letters with their half-lives shown below the nuclide symbol; radionuclides that are not part of this discussion are left blank. The s-process pathway comes through monoisotopic 181Ta, branches first at 185W, and then again at 186Re. Both branches combine at 188Os. Three processes occur at 186Re: β decay to 186Os (92%), electron capture to 186W (8%, lower at stellar temperatures), and neutron capture to 187Re. The r process is shown as oblique long arrows emerging from the lower right corner of the field and terminating at the first stable nuclide. Isotopes shielded from the r process include 186Os and 187Os; 184Os is a p-only isotope; 192Os is an r-only isotope (s-process contribution «1% relative). Z, atomic number; N, neutron number.

The measured Os isotope data (9) are displayed in epsilon units (Fig. 2), where ϵOs is 104 parts deviation from the mean of H-group ordinary chondrites measured in this study. This is because of the overlap of the H-group Os isotope ratios with those of the terrestrial materials, which are interpreted as bulk solar values (Fig. 2). The ϵ186 values are corrected for 190Pt decay over 4.558 billion years to initial values (ϵ186i; Fig. 2 and table S1). This results in a systematic ∼1.1-epsilon correction for all chondrite samples. Because 189Os and 192Os are used for correcting for instrumental mass fractionation, all data have ϵ189 and ϵ192 of zero. The 189Os and 192Os isotopes were chosen for normalization because in solar system Os, these isotopes contain the smallest contribution of s-process nucleosynthesis [4.8% and 0.4%, respectively (68)]. This normalization best illustrates variations in the s-process components relative to those in r-process components (10).

Fig. 2.

(A) Normalized Os isotopic compositions of H-group (diamonds) and LL-group (circles) ordinary chondrites. (B) Normalized Os isotopic compositions of enstatite chondrites. Open triangles, EL6; solid triangles, Indarch EH4. (C) Normalized Os isotopic compositions of the carbonaceous chondrites Allende (open squares), Ornans (half-solid squares), and Tagish Lake (TL) (solid squares). The 187Os data are not shown because of large variable contributions from radioactive decay (9). The shaded regions are the ±2σ error envelope on 39 runs of the terrestrial standard (external precision). Terrestrial mantle samples (table S1) all plot within these envelopes. Individual data points for each sample have ±0.25 ϵOs (2σ) or better uncertainties. The calculated atomic percentage of s and r processes for each isotope listed are based on models from Beer et al. (8).

Tagish Lake (C2), and to a lesser extent Ornans (CO3), display deficiencies in the s-process isotopes relative to most of the ordinary and enstatite chondrites measured, with anomalies in the order: 186Os > 188Os > 190Os (Fig. 2). These deficiencies correlate with the s-process abundances of each isotope. Also, Parnellee (LL3.6) and Indarch (EH4) exhibit small negative ϵ186i (98.9% s-process isotope) anomalies.

The variations observed in ϵOs could result from cosmic-ray exposure effects, incomplete digestion of insoluble Os carrying phases, or inhomogeneous distribution of the s- and r-process isotopes in the solar nebula. The effects on Os isotopes from cosmic-ray exposure can be estimated (11). For an exposure age of 100 million years, the effect should be from 0.2 to 1 epsilon units for burnout and production of the different Os isotopes. The exposure age for Tagish Lake calculated from 21Ne is 7.8 million years (12), and hence, the effects for cosmic-ray exposure should be negligible.

Osmium is hosted in Fe-Ni alloy, refractory metal nuggets, and possibly in presolar SiC or graphite, similar to Ru (1315). To determine whether incomplete digestion of an Os carrier was responsible for the effect, we performed a digestion of Tagish Lake using sealed metal jackets with CO2 overpressure, at temperatures of 325°C for 1 week (9, 16), compared with no overpressure, 240° to 270°C, and 48 to 72 hours for the other digestions. The Os isotopic data for this digestion retains the Os isotopic anomalies (Fig. 2) but not to the same extent as the other digestions for this rock. In addition, the four separate digestions of different clasts from the Tagish Lake meteorite show variable anomalies with systematic depletions, where 186Os > 188Os > 190Os. These observations imply that there is an s-process deficit in Os extracted from Tagish Lake and suggest that whatever phase is carrying the s-process Os, it is either heterogeneously distributed or variably accessed by the digestions. From the s-process anomalies observed in Tagish Lake, it can be estimated that 30 to 50 parts per million (ppm) of the total Os (14 to 24 pg/g out of 476 ng/g) are located in the s-process carrier. If the carrier phase is presolar SiC, present at ∼5 to 8 ppm by weight of Tagish Lake (17), then the s-process Os concentration is ∼1 to 5 ppm. The actual Os concentration could be higher because isotopically normal Os is incorporated from the stellar atmosphere. Elemental abundances in presolar SiC by synchotron x-ray fluorescence for Ru (2 to 57 ppm) and Mo (1 to 54 ppm) (15) are of the same order of magnitude as the mean Os abundance (5 to 8 ppm, if SiC is assumed as the carrier phase) in the s-process carrier inferred here. The chemical similarity of Os to Ru indicates that some Os may occur in SiC, even though neither element is known to form carbides. Osmium occurs as refractory metal inclusions within presolar graphite grains (14), raising the prospect that SiC acts as a protective coating (against acid attack) for ultrarefractory metal condensates. Graphite is soluble in nitric acid, and Os alloys are expected to dissolve in aqua regia, so these grains are not the likely carriers of the observed anomaly given that such grains would have been consumed in the digestions carried out here, unless shielded in a resistant phase such as SiC.

Chondrite metamorphism has been documented to destroy presolar graphite and SiC (18). SiC typically has not survived to higher metamorphic grades of >3.8 for ordinary chondrites, and ∼5 for enstatite chondrites (19, 20) consistent with such samples showing no Os isotope anomalies (Fig. 2). Thus, the likely interpretation is that an undigested carrier phase, probably SiC, retains the presolar s-process Os required to yield the solar s/r ratio for Os exhibited by H-group ordinary chondrites and terrestrial samples (Fig. 2). This is supported by the observation that two chondrites that underwent less metamorphism—Parnellee (LL3.6) and Indarch (EH4)—show s-process anomalies in their Os isotope profiles, but less so than Tagish Lake (C2) and Ornans (CO3). Presolar SiC is present in Indarch and some unequilibrated ordinary chondrites (19, 20). Sharps (H3.4) and Chainpur (LL3.6) do not exhibit Os isotope anomalies but instead plot within the range of epsilon values exhibited by the other H-group ordinary chondrites in Fig. 2.

Although most of Os originates in the r process (90%), about 10% of the solar system Os is produced by the s process (68). Correlated trends between ϵ186i, ϵ188, and ϵ190 can be used to evaluate the nature of the anomalous Os isotopic component (Fig. 3). Mixing lines between average s-process Os isotopic composition (8) and average solar Os (i.e., ϵ186i, ϵ188, and ϵ190 = 0) are resolvable from the correlations of the chondrite data for ϵ188 to ϵ186i and ϵ190 to ϵ186i but match the ϵ188 to ϵ190 correlation within error of the measurements (Fig. 3).

Fig. 3.

(A and B) ϵ190 versus ϵ188 and (C and D) ϵ186i versus ϵ188 for carbonaceous chondrites (solid squares), ordinary chondrites (open diamonds), and enstatite chondrites (open triangles). Solar values are at zero. The solid lines are for a regression between solar values at zero and three model solar s-process (7, 8) values, which are solid diamonds plotted at times 10–4 of their respective values. In (B) and (D) the ±2σ error bars are shown based on replicate measurements of the terrestrial standard (external precision). The dashed lines are regressions between the Tagish Lake data and solar values at zero. Any component not accessed in the Tagish Lake digestions should lie along these lines at positive values. The Tagish Lake regression line agrees within error of the solar s-process regression line for ϵ188 versus ϵ190. However, the Tagish Lake regression line for ϵ188 versus ϵ186i is resolvable from the solar s-process regression line and trends toward positive values with a depletion in ϵ186i at a given ϵ188, relative to the solar regression line.

This implies that the 186Os/188Os ratio of the anomalous s-process component was lower relative to average solar system s-process Os. Here, we term this insoluble Os component Os(i). There are two branches in the s-process pathway at 185W and 186Re that allow the s process to flow to 186W, 187Re, and then to 188Os (Fig. 1). Neutron capture on 185Re forms 186Re that decays 92% of the time by β (half-life of 3.8 days) to 186Os. The two branch points are relatively insensitive to stellar temperature but sensitive to stellar neutron density. These branch points have been used to estimate that the neutron density for average solar s process is 4 × 108 n/cm3 (5, 21). The Os(i) fraction of the solar s process requires a stellar source with higher neutron density. The neutron density (nn) of the source was modeled by solving the simultaneous linear equations describing s-process flow with β decay branching to yield Math(1) Math Math where vT is the thermal neutron velocity, λ is the decay constant, Ns is the s-process abundance, 〈σ〉 is the maxwellian-averaged neutron capture cross-section, and τ0 is the average neutron exposure (21). The branching decay of 186Re to 186W by electron capture was neglected for this calculation, because electron capture is diminished at the higher degree of ionization prevailing in stellar interiors (22). Using cross-sections, decay constants, and τ0 typical of the s process (5, 21, 22), we calculated 186Os/188Os ratios as a function of stellar neutron density and temperature (Fig. 4). The calculated neutron density for solar s process is similar to that obtained in (21). The 186Os/188Os ∼0.48 in Tagish Lake Os(i) is matched at stellar neutron densities of 6 × 108 to 10 × 108 n/cm3.

Fig. 4.

The s-process 186Os/188Os ratio depicted as a function of stellar neutron density. The input data, including decay constants and n-capture cross-sections, were adjusted for stellar temperature and the curves are shown for T = 1 × 108 K (solid black curve) and for T = 3 × 108 K (dashed curve). The average solar s-process 186Os/188Os (7, 8) is shown as the gray shaded band, the r-process 186Os/188Os ∼0, and the Tagish Lake Os(i) component has 186Os/188Os ∼0.48. The range of neutron density implied by the model for the average solar s process is shown by the double-arrow compared with the result from Käppeler et al. (21), which is shown as an error bar. The range of neutron density required to explain the missing component in Tagish Lake is about 2 to 4 times as high as the solar value.

There are two major neutron sources for the s process produced by helium-burning stellar reactions (α capture): 13C(α,n)16O and 22Ne(α,n)25Mg. The 13C-source operates at T ∼1 × 108 K and the 22Ne source operates at T ∼3 × 108 K (8, 23). Coupled models of stellar evolution and nucleosynthesis in low-mass asymptotic giant branch stars indicate that during a double neutron pulse the 22Ne source contributes importantly to the abundances of nuclides produced at high neutron densities (up to 1010 n/cm3) (23). This has been identified with the meteoritic Ne-E(H) component (23) and may also contribute the Os(i) component in Tagish Lake. Likely, the main component of s-process Os (and other heavy elements) may be resolved to be a mixture of varied neutron density sources operating in multiple low-mass stars with variable metallicity. Our technique resolved solar s-process Os in primitive chondrites into two components: Os(i) that is likely trapped in SiC grains, and aqua regia–soluble Os that is hosted by other phases (magnetite or Fe sulfides), which provide insights into the galactic chemical evolution of Os. The extractable Os either condensed into acid-soluble minerals from stellar outflows with C/O < 1, or it initially condensed in SiC grains which were then selectively destroyed in the interstellar medium (possibly older grains).

This interpretation—that the anomalous Os isotopic composition in unequilibrated chondrites results from incomplete access of up to 50 ppm of the total Os present during digestion—has important ramifications to understanding the presence or absence of isotopic anomalies in bulk meteorites for other elements, including Zr, Mo, and Ru (4, 2429). Because the Os abundance in bulk meteorites is small (ϵ1 ppm) and because Os is one of the first elements to condense, mixing within the solar nebula before condensation into planetesimals must have been extremely efficient in order to result in a homogeneous Os isotopic composition of ±0.25 epsilon units (i.e., ±25 ppm) observed in chondrites with greater metamorphic equilibration.

Supporting Online Material

Materials and Methods

Figs. S1 and S2

Tables S1 to S4


References and Notes

Stay Connected to Science

Navigate This Article