Experimentally determined mineral/melt partitioning of first-row transition elements (FRTE) during partial melting of peridotite at 3 GPa

Ratios of first-row transition elements (FRTE), such as Fe/Mn and Zn/Fe, may be fractionated differently by partial melting of peridotite than by partial melting of recycled lithologies like eclogite, and therefore may be useful as indicators of the source lithologies of mantle- derived basalts. Interpretation of basalt source lithologies from FRTE ratios requires accurate assessment of FRTE partitioning behavior between peridotitic minerals and coexisting melts. We present experimental determinations of partition coefficients for several of the FRTE (Sc, Ti, V, Cr, Mn, Fe, Co, Zn) and Ga and Ge between basaltic melt and olivine, garnet, pyroxenes, and spinel at 3 GPa. Because mineral/melt partitioning is sensitive to phase compositions, a key feature of these experiments is that the melts and minerals are known from previous experiments to be in equilibrium at the solidus of garnet peridotite at 3 GPa. Therefore, these partition coefficients are directly applicable to near-solidus partial melting of the mantle at 3 GPa. We use these partition coefficients to calculate compositions of model partial melts of peridotite and compare these to natural OIB. Model partial melts of peridotite have lower Fe/Mn (<62) and higher Co/Fe (>7 * 10(-4)) than many primitive OIB, which implies that some other source lithology participates in the formation of many OIB. Alternatively, these ratios may potentially be produced by garnet peridotite if the source contains similar to 0.3% Fe2O3, consistent with observations from continental xenoliths. Zn/Fe is a less sensitive indicator of non-peridotite source lithology than either Fe/Mn or Co/Fe, as Zn/Fe in partial melts of peridotite overlaps with >75% of primitive OIB. Ga and Sc are fractionated significantly by residual garnet, and high Ga/Sc may indicate the presence of garnet in basalt source regions. When taking into account several FRTE ratios simultaneously, few OIB appear to be consistent with derivation solely from a reduced peridotitic source. The source either must have a modest non-peridotitic component, be Fe-enriched, or be slightly oxidized. (C) 2012 Elsevier Ltd. All rights reserved.