ELEMENT PARTITIONING BETWEEN MG-PEROVSKITE, MAGNESIOWUSTITE, AND SILICATE MELT AT CONDITIONS OF THE EARTHS MANTLE

Magnesium-perovskite and magnesiowustite are the dominant phases in the lower mantle of the Earth. We have determined experimentally element partitioning behavior between Mg-perovskite and silicate melt and between magnesiowustite and silicate melt at 24.5 (+/-0.5) GPa and 2400 (+/-100)-degrees-C using a mantle composition (KLB-1) doped with trace elements. Average Mg-perovskite/melt partition coefficients and 2 sigma uncertainty estimates from three experiments are: Mg = 1.0 (+/- 0.1), Al = 1.2 (+/-0.3), Si = 1.17 (+/-0.05), Ca = 0.3 (+/-0.3), Ti = 1.3 (+/-0.3). Average magnesiowustite/melt partition coefficients and 2 sigma uncertainty estimates from two experiments are: Mg = 2.2 (+/-0.2), Al = 0.36 (+/-0.07), Si = 0.005 (+/-0.005), Ca = 0.01 (+/-0.02). Titanium is below detection. Trace element partition coefficients from a sample doped with V, Cr, and Mn are: Mg-perovskite-V = 1.14 (+/-0.06), Cr = 1.1 (+/-0.1), Mn = 0.62 (+/-0.08); magnesiowustite-V = 0.45 (+/-0.01), Cr = 1.27 (+/-0.03), Mn = 0.77 (+/-0.06). The ratios of partition coefficients for some of the refractory and moderately refractory lithophile elements listed above deviate significantly from unity. If Mg-perovskite or Mg-perovskite and magnesiowustite segregated from a terrestrial magma ocean, ratios of these elements (Al/Ti, Ca/Ti, Ca/Al, Mg/Si, Si/Al) would have been imparted upon the primitive mantle of the Earth which are inconsistent with inferred values based on analyses of naturally occurring samples. Olivine addition into the upper mantle can be invoked to eliminate this constraint for some elements (i.e., Mg/Si), but will exacerbate the problem for other elements (i.e., Si/Ca, Si/Al). If magnesiowustite segregated alone, ratios of these elements (Mg/Al, Mg/Ca, Si/Ca, Mg/Si, Si/Al) would be inconsistent with inferred values. There is no surviving evidence of mineral fractionation implying that either (a) the Earth was never substantially molten at the end of accretion; (b) segregation of minerals from magma was suppressed by vigorous convection or some other mechanism; or (c) evidence of fractionation was subsequently destroyed by solid-state mantle convection. The abundances of V, Cr, and Mn in the mantles of the Earth and Moon are very similar to each other, and differ from all other sampled differentiated plantary bodies. The origin of this similarity is currently unknown, but is not attributable to differentiation of metallic or silicate phases at low pressures. Our results show that this abundance pattern cannot result from high pressure and temperature fractionation of Mg-perovskite and/or magnesiowustite at conditions relevant to the uppermost lower mantle of the Earth.