Re-Os Isotopic Evolution of the Earth's Mantle,
Archean to Present

S. B. Shirey Carnegie Institution of Washington, 5241 Broad Branch Rd., NW, Washington, DC 20015, USA

shirey@dtm.ciw.edu

R. J. Walker Dept. of Geology, University of Maryland, College Park, MD 20742, USA

rw67@umail.umd.edu

Among the long-lived radiogenic isotope tracers, the Re-Os system is different in that Re and Os are both chalcophile and siderophile in magmas. The growing body of Os isotopic data for relatively young samples has broadened the understanding of existing mantle reservoirs. Upper mantle Os isotopic composition was set by the late accretion of chondritic materials as is reflected by the source of MORB having an approximately chondritic Os isotopic composition in spite of the possibility of long-term depletion of the MORB source in Re by melting or enrichment by recycling of crustal Re. Peridotite xenoliths of the subcontinental lithospheric mantle carried by relatively recent (0.1-0.35 Ga) kimberlitic/alkalic volcanism penetrating most Archean cratons typically show subchondritic 187Os/188Os ratios that can only be explained by long-term evolution with a low Re/Os which is a likely result of ancient melt extraction events. In contrast, many oceanic island basalts have suprachondritic Os isotopic compositions that require sources with long-term higher-than-chondritic Re/Os. In some instances, these enrichments are consistent with current models that explain their lithophile element isotopic systematics by recycling of various oceanic or continental crustal components. In other instances, such models may be inadequate.

Although the heterogeneous nature of Os isotopes in the convecting mantle is now well known, it is currently not known when the Os isotopic compositions of these mantle reservoirs began to diverge. The Re-Os systematics of ancient rocks, therefore, have the potential to provide a unique view of Precambrian mantle evolution. The evolution of Os in the Precambrian mantle can be best deciphered
via the careful study of ultramafic rocks such as komatiites, basaltic komatiites, picrites, and ultramafic portions of
ophiolites. Most komatiitic and picritic rocks likely derive from "deep" mantle sources whereas ophiolites may provide information about the upper oceanic mantle. Such rocks generally have low 187Re/188Os (<10), suitable for obtaining precise initial melt composition information, and Os abundances (>1 ppb) that are too high to be significantly affected by modest crustal contamination or seawater interaction. We have found, however, that many whole rock Re-Os systems (especially those containing sulfides) are very sensitive to metamorphic open-system behavior. Consequently, Re-Os isotopic studies of ancient rocks will likely require the analysis of separated primary phases such as chromite, olivine and pyroxene. Preliminary studies of these phases show encouraging results for deciphering the primary magmatic compositions of Os in ancient rocks.

Very little reliable Re-Os data exist for well-characterized Precambrian ultramafic systems. Whole-rock chromite-rich rocks from the Ruth Well komatiites, Pilbara craton, Australia (ca. 3.0-3.3 Ga) give an initial gOs of -0.1± 1.5. Similarly, gOs values for chromites from the ultramafic portion of the Stillwater Complex, Montana (ca. 2.7 Ga) are highly variable, but range to as low as +1.8. Clearly, "chondritic" Os reservoirs existed in the mantle during the Archean. Early Proterozoic ferropicritic volcanism at Pechenga, Russia (ca. 1.98 Ga) had initial gOs of -2.0 ± 1.0, while chromitites from the Outokumpu ophiolite, Finland (ca. 1.97 Ga) had initial gOs of -0.5+ 1.0. Basaltic komatiites from the Cape Smith foldbelt (ca. 1.9 Ga) had inital gOs of
-2.0+ 3.0. Again, it is clear that "chondritic" mantle reservoirs existed during the Early Proterozoic. There is at present no evidence for significantly supra- or sub-chondritic Os in any Early Proterozic or older mantle reservoir.

In contrast to the older rocks, Late Proterozoic Keweenawan picrites from the Midcontinent Rift, North America (ca. 1.1 Ga) have significantly higher gOs values averaging about +10. Indeed, this enrichment in plume-derived sources appears to consistently follow through the Phanerozoic to the present. Ultramafic samples from the Noril'sk area of the Siberian flood basalt province (ca. 250 Ma) have gOs values ranging from +6 to +14, and values
for komatiites from Gorgona Island (ca. 89 Ma) range from -1.1 to +13.7. Thus, it appears that older ultramafic rocks have initial Os isotopic compositions that are generally "chondritic", but younger rocks, presumably derived from plume sources show strong evidence for long term Re enrichment during the Late Proterozoic and throughout the Phanerozoic just as many modern OIB sources do.

The limited existing database for Precambrian rocks provides no evidence for an early mantle that was stratified with respect to highly siderophile elements. If this conclusion holds up under further scrutiny, the implication is that the highly siderophile elements added to the mantle via late accretion were quickly mixed throughout the portions
of the mantle (upper and lower?) from which the studied ultramafic rocks were derived. Increasing incorporation
of recycled oceanic crust, or interaction between the mantle and the core led to enrichments of 187Os in some mantle reservoirs probably starting during the Late Proterozoic.