| Summary: | Young (61 Ma) unaltered picrites from Baffin Island, northeast Canada, possess some of the highest ³He/⁴He (up to 50 Ra) seen on Earth, and provide a unique opportunity to study primordial mantle that has escaped subsequent chemical modification. These high-degree partial melts also record anomalously high ¹⁸²W/¹⁸⁴W ratios, but their Sr-Nd-Hf-Pb isotopic compositions (including ¹⁴²Nd) are indistinguishable from those of North Atlantic mid-ocean ridge basalts. New high precision Fe and Zn stable isotope analyses of Baffin Island picrites show limited variability with δ⁵⁶Fe ranging from −0.03‰ to 0.13‰ and δ⁶⁶Zn varying from 0.18‰ to 0.28‰. However, a clear inflection is seen in both sets of isotope data around the composition of the parental melt (MgO = 21 wt%; δ⁵⁶Fe = 0.08 ± 0.04‰; and δ⁶⁶Zn = 0.24 ± 0.03‰), with two diverging trends interpreted to reflect the crystallisation of olivine and spinel in low-MgO samples and the accumulation of olivine at higher MgO. Olivine mineral separates are significantly isotopically lighter than their corresponding whole rocks (δ⁵⁶Fe ≥ −0.62‰ and δ⁶⁶Zn ≥ −0.22‰), with analyses of individual olivine phenocrysts having extremely variable Fe isotope compositions (δ⁵⁶Fe = −0.01‰ to −0.80‰). By carrying out modelling in three-isotope space, we show that the very negative Fe isotope compositions of olivine phenocryst are the result of kinetic isotope fractionation from disequilibrium diffusional processes. An excellent correlation is observed between δ⁵⁶Fe and δ⁶⁶Zn, demonstrating that Zn isotopes are fractionated by the same processes as Fe in simple systems dominated by magmatic olivine. The incompatible behaviour of Cu during magmatic evolution is consistent with the sulfide-undersaturated nature of these melts. Consequently Zn behaves as a purely lithophile element, and estimates of the bulk Earth Zn isotope composition based on Baffin Island should therefore be robust. The ancient undegassed lower mantle sampled at Baffin Island possesses a δ⁵⁶Fe value that is within error of ...
|
|---|