Zinc isotope fractionation in mantle rocks and minerals, and a revised δ66Zn value for the Bulk Silicate Earth
Résumé
The isotope composition of zinc (Zn) can be an important tracer of planetary formation and differentiation, but estimates of the Bulk Silicate Earth (BSE) and the extent and origin of Zn isotope variations in the mantle continue to be debated. We report 153 high-precision Zn isotope analyses of whole-rock samples and minerals of well-characterized, mantle-derived spinel and garnet peridotites from five sites in Siberia, Mongolia and China: 38 xenoliths in basaltic rocks (fertile to strongly melt-depleted as well as variably metasomatized or veined peridotites) and 3 orogenic peridotites. Data quality was rigorously tested by twenty-two duplicated analyses using several dissolution techniques. Spinel (Sp) contains 760-1460 μg/g Zn and hosts up to 58 % of whole-rock Zn budget in the peridotites followed by olivine (Ol, 30-66 % of whole-rock Zn budget), orthopyroxene (Opx), clinopyroxene (Cpx) and garnet, with abundance ratios of [Zn]Sp/[Zn]Ol = 15-47, [Zn]Ol/[Zn]Opx = ∼1.5, [Zn]Ol/[Zn]Cpx = ∼6. Spinel has higher δ66Zn values than coexisting silicates (∼0.12 ‰ higher than in Ol). Minerals in orogenic peridotites show disequilibrium Zn element and isotope partitioning due to slow inter-grain Zn diffusion at retrograde metamorphism. Fertile peridotites (≥3 wt% Al2O3) contain 52-60 μg/g Zn, usually higher than melt-depleted (<3 wt% Al2O3) rocks (40-53 μg/g). The δ66Zn values in the majority of spinel and garnet peridotites range from 0.11 to 0.26 ‰ and show no systematic co-variations with melt extraction indices (e.g., Al2O3, Mg#), suggesting, unlike some previous studies, that partial melting does not appreciably affect Zn isotope composition of residual mantle. Garnet lherzolites yield an average δ66Zn of 0.22 ± 0.09 ‰ (2SD), similar to that (0.18 ± 0.07 ‰, 2SD) for shallower spinel peridotites, indicating uniform Zn isotope composition with depth. Four fertile lherzolites from earlier work, duplicated in this study to assess inter-laboratory bias, fall in the same δ66Zn range, suggesting that previously reported higher δ66Zn for these samples could be due to analytical problems. We thus estimate the δ66Zn of the BSE at 0.19 ± 0.08 ‰ (2SD) based on suitable fertile peridotites from several localities in this study and the literature. The δ66Zn values (-0.10 to 0.04 ‰) are much lower in Fe-rich (9.3-13.1 wt% FeOT) and veined spinel peridotites as well as in spinel pyroxenite veins from Mongolia (-0.20 to 0.11 ‰). The Fe-rich peridotites and pyroxenites are rich in Zn (up to 113 μg/g) and were previously found to have anomalously low δ56Fe (-0.33 to - 0.27 ‰) and δ53Cr (-1.36 to - 0.05 ‰). We attribute these features to kinetic isotope fractionation during reaction with mafic melts migrating in conduits in the lithospheric mantle or intrusion of Fe-Zn-rich melts with light Fe-Zn isotope compositions. Kinetic Zn isotope fractionation, rather than equilibrium partial melting, may be a major cause of Zn isotope heterogeneity in the mantle.