Partial melting and melt percolation in the mantle: The message from Fe isotopes

Abstract

High precision Fe isotope measurements have been performed on various mantle peridotites (fertile lherzolites, harzburgites, metasomatised Fe-enriched peridotites) and volcanic rocks (mainly oceanic basalts) from different localities and tectonic settings. The peridotites yield an average δ 56Fe = 0.01‰ and are significantly lighter than the basalts (average δ 56Fe = 0.11‰). Furthermore, the peridotites display a negative correlation of δ 56Fe with Mg# indicating a link between δ 56Fe and degrees of melt extraction. Taken together, these findings imply that Fe isotopes fractionate during partial melting, with heavy isotopes preferentially entering the melt. The slope of depletion trends (δ 56Fe versus Mg#) of the peridotites was used to model Fe isotope fractionation during partial melting, resulting in α mantle-melt ≈ 1.0001–1.0003 or ln α mantle-melt ≈ 0.1–0.3‰. In contrast to most other peridotites investigated in this study, spinel lherzolites and harzburgites from three localities (Horoman, Kamchatka and Lherz) are virtually unaffected by metasomatism. These three sites display a particularly good correlation and define an isotope fractionation factor of ln α mantle-melt ≈ 0.3‰. This modelled value implies Fe isotope fractionation between residual mantle and mantle-derived melts corresponding to Δ 56Fe mantle–basalt ≈ 0.2–0.3‰, i.e. significantly higher than the observed difference between averages for all the peridotites and the basalts in this study (corresponding to Δ 56Fe mantle–basalt ≈ 0.1‰). Either disequilibrium melting increased the modelled α mantle-melt for these particular sites or the difference between average peridotite and basalt may be reduced by partial re-equilibration between the isotopically heavy basalts and the isotopically light depleted lithospheric mantle during melt ascent. The slope of the weaker δ 56Fe–Mg# trend defined by the combined set of all mantle peridotites from this study is more consistent with the generally observed difference between peridotites and basalts; this slope was used here to estimate the Fe isotope composition of the fertile upper mantle (at Mg# = 0.894, δ 56Fe ≈ 0.02 ± 0.03‰). Besides partial melting, the Fe isotope composition of mantle peridotites can also be significantly modified by metasomatic events, e.g. melt percolation. At two localities (Tok, Siberia and Tariat, Mongolia) δ 56Fe correlates with iron contents of the peridotites, which was increased from about 8% to up to 14.5% FeO by post-melting melt percolation. This process produced a range of Fe isotope compositions in the percolation columns, from extremely light (δ 56Fe = − 0.42‰) to heavy (δ 56Fe = + 0.17‰). We propose reaction with isotopically heavy melts and diffusion (enrichment of light Fe isotopes) as the most likely processes that produced the large isotope variations at these sites. Thus, Fe isotopes might be used as a sensitive tracer to identify such metasomatic processes in the mantle.