Si isotope composition of Earth’s mantle and oceanic basalts

We have re-examined mass dependent silicon isotope variations in the terrestrial mantle by exploiting the capability of critical mixture double spiked analyses to measure Si isotope ratios with a reproducibility close to analytical precision (±0.03 ‰ on δ^30/28Si). Our measurements are sufficiently precise to determine Si isotopic differences between mantle phases (Δ^30/28Si_Ol/Cpx = 0.075 ± 0.037 ‰, Δ^30/28Si_Ol/Opx = 0.056 ± 0.028 ‰, and Δ^30/28Si_Ol/Gar of 0.123 ± 0.075 ‰) and crucially between melt and olivine (Δ^30/28Si_Ol/Melt of −0.001 ± 0.029 ‰). Applying these fractionation factors, we predict insignificant (<0.03 ‰) Si isotope variations during basaltic melt differentiation, but detectable fractionation during mantle partial melting; typical melts of a peridotite are ∼0.03 ‰ isotopically heavier than their source, and melts of pyroxenite are ∼0.04 to ∼0.1 ‰ isotopically heavier than their sources. These calculations give context for our δ^30/28Si analyses of various samples of the mantle, provided by mantle peridotites, mid-ocean ridge basalts (MORB) and ocean island basalts (OIB). Eight peridotite measurements, chosen to cover a range of compositions and settings, have a uniform Si isotope composition of −0.308 ± 0.013 ‰ (2 s.e., δ^30/28Si). Twenty-one MORB analyses, from fast to ultra-slow spreading ridges from three ocean basins, and seventeen OIB samples, that span much of the range of their global radiogenic isotope ratios, yield similar mean values of −0.280 ± 0.009 ‰ (2 s.e., δ^30/28Si) and −0.271 ± 0.013 ‰ (2 s.e., δ^30/28Si), respectively. The absence of systematic δ^30/28Si variability in MORB from ridges with variable spreading rates, in contrast to recent observations of mass dependent Mg and Fe isotope fractionation, argues against kinetic fractionation of Si isotopes during melt transport. Moreover, our combined dataset shows that the differences between mantle and mantle derived melts are in good agreement with our calculated values for equilibrium melting, implying δ^30/28Si homogeneity in diverse samples of the accessible mantle at the level of ∼±0.015 ‰. The modelled difference in Si isotope composition of peridotite melt and pyroxenite melt furthermore allows us to quantify pyroxenite melt contribution to oceanic basalt. The modelled difference in δ^30/28Si between peridotite melt and pyroxenite melt combined with the statistically similar mean δ^30/28Si values for MORB and OIB indicate an upper limit of pyroxenite melt contribution of ∼18 % to MORB samples and ∼23 % to OIB samples on average.