Osmium isotope constraints on formation and refertilization of the non-cratonic continental mantle lithosphere

Abstract

Listen

Translate article

It is increasingly accepted that many lherzolites in the lithospheric mantle formed by refertilization of depleted harzburgites, rather than by small degrees of melt extraction from fertile precursors. Such refertilization is often assumed to be linked to tectonic events unrelated to and much younger than the original partial melting episodes that left the depleted harzburgitic residues. Osmium isotopes provide a means of testing this assumption. Data for 187Os/188Os are now available from more than 50 studies of peridotites from the non-cratonic lithospheric mantle, including both xenoliths in alkali basalts and mantle tectonites of subcontinental affinity. The results from these studies confirm the long-recognized observation that whole rock Os isotope compositions are frequently correlated with indices of peridotite fertility such as Al2O3 content, with >50% of the samples having 187Os/188Os ratios plotting within ±1.5% of correlation lines between the estimated Primitive Upper Mantle (PUM) composition and the least radiogenic sample in each suite. A simple melt percolation model was developed that shows that such trends cannot result from direct precipitation of exsolved radiogenic sulfides, because any Os initially present in the circulating magma would be trapped by sulfides at the very bottom of the column, due to the extreme chalcophility of this element. Refertilized peridotites higher in the column would thus have unradiogenic Os compositions. The absence of fertile lherzolites with unradiogenic Os isotope compositions, the expected products of percolation of young melts through ancient harzburgites, argues that this process is extremely rare. Instead, the addition of Re during melt percolation occuring soon after harzburgite depletion would produce correlations between Re/Os and Al2O3 that with time would develop into correlations between 187Os/188Os and Al2O3 passing through the PUM composition. Such correlations would be linear, as observed, if DResulf/sil is ~400, the approximate value predicted for this parameter for FeO-rich magmas at fO2 and fS2 values appropriate for non-arc mantle melting conditions. This perspective is fully consistent with the substantial variation in 187Os/188Os ratios found among sulfides taken from individual hand samples, if radiogenic ingrowth over time is associated with extensive, but incomplete, diffusive reequilibration.The conclusion that refertilization must follow soon after harzburgite formation has several implications. First, it suggests that the original melt depletion episodes are causally linked to the subsequent refertilization, which could result from percolation of the remaining melts through the cooling lithosphere. Second, early refertilization has consequences for the long-term stability of the lithosphere, since lherzolites are denser rocks than harzburgites. A final observation is that the systematic correlations between Os isotope compositions and fertility indices observed in most non-cratonic peridotite suites are unlikely to form in suprasubduction settings, where melts often carry ultramafic xenoliths with radiogenic and chaotic 187Os/188Os ratios. Instead, extensive melting above a mantle plume, followed soon after by refertilization with remaining melts as the plume cools, could explain the observed systematics. This newly formed mantle lithosphere could then be tectonically juxtaposed beneath continental crust formed above subduction zones.