Zinc isotope fractionation between Cr-spinel and olivine and its implications for chromite crystallization during magma differentiation

Abstract

We report the first zinc isotope data (expressed as δ66Zn relative to the JMC-Lyon standard) for chromian spinels (Cr-spinels) and coexisting olivines in oceanic peridotites. All spinel-olivine pairs fall on the 1:1 fractionation line in the diagram of δ66Znspinel versus δ66Znolivine, suggesting equilibrium isotope fractionation. Cr-spinels are always isotopically lighter than coexisting olivines (Δ66Znspinel-olivine = δ66Znspinel − δ66Znolivine = −0.50‰ to −0.33‰; n = 13), which is opposite to the positive Zn isotope fractionation between Al-spinel and olivine observed in cratonic peridotites. The “inverse” Zn isotope fractionation between Cr-spinel and olivine is unlikely to have been caused by low-temperature alteration of the oceanic peridotites, given the lack of correlation between Δ66Znspinel-olivine values and chemical indices of serpentinization and weathering (e.g., LOIs, MgO/SiO2). Ionic model calculation indicates that even considering that Zn occurs at both tetrahedral and octahedral sites in the crystal lattice of spinel, the magnitude of equilibrium fractionation is still inconsistent with the observed Zn isotope fractionation between Cr-spinel and olivine. We suggest a “chemical effect” in which the ZnO bond length (tetrahedral site) in spinel increases when Cr substitutes Al in octahedral site, which is corroborated by the striking negative correlation of δ66Zn with Cr# [molar Cr3+/(Cr3+ + Al)] in natural spinels. Therefore, Zn isotopic compositions of natural spinels can be highly variable depending on their chemical compositions.During magma differentiation, zinc is moderately incompatible in silicate minerals (olivine and pyroxene) but highly compatible in Cr-spinels/chromites that have Zn contents tens of times higher than those of basaltic melts. Given its light Zn isotopic composition, chromite crystallization–if any–can evidently elevate δ66Zn and lower Zn contents of the residual melts. Lunar mare basalts are typically characterized by higher δ66Zn and lower Zn contents relative to terrestrial basalts, but modelling suggests that chromite crystallization during magmatic differentiation is unlikely to account for Zn isotopic and elemental data of lunar mare basalts. Global ocean island basalts (OIBs) and some intraplate alkali basalts have systematically higher δ66Zn and higher Zn contents than those of the normal mantle-derived melts (e.g., mid-ocean ridge basalts; MORBs), which contradicts with a chromite crystallization model. Instead, such signatures can reflect recycling of surface carbonates into the Earth’s deep mantle, reinforcing the application of zinc isotopes as a tracer of deep carbonate recycling.