Abstract

Mantle melting that produces ocean island basalts takes place at temperatures and pressures significantly higher than the conditions at which they erupt or are intruded in the crust/shallow upper mantle. To the degree that the olivine-liquid nickel partition coefficient, D_Ni , depends on temperature and pressure, it is important that models used to describe Ni partitioning during mantle melting include data from experiments at elevated temperature and pressure. Available data on Ni partitioning is dominated by 1-atm experiments in which temperature and liquid composition are highly correlated, making it difficult to separate the effects of these variables on the observed variations in D_Ni based on 1-atm experiments alone.

We conducted experiments on a mixture of MORB and olivine at 1 atm (1400°C) and 1-3 GPa (1450-1550°C). We present data from a series of reversed experiments where temperature and pressure were increased in such a way that the liquid composition remained approximately constant (MgO about 17 wt%), effectively isolating the effects of temperature and pressure from those of liquid composition on D_Ni. The resulting partition coefficient decreases from ~5 to 3.8 (by wt) as the temperature increases from 1400 to 1550°C, and is fit well by a simple thermodynamic expression where D_Ni is a function of temperature, olivine, and liquid composition.

Based on a simple model in which partial melting to produce a primary liquid and subsequent crystallization of that liquid take place at two different pressures and temperatures, the temperature dependence of D_Ni predicts the crystallization of low-pressure olivine phenocrysts with higher NiO contents than the olivines in the source region. Observed NiO concentrations of magnesian olivine phenocrysts from Mauna Loa, Mauna Kea and Kilauea are consistent with derivation from primary partial melts produced at the base of the lithosphere from peridotites whose olivines have a distribution of NiO concentrations equal to that observed in spinel peridotites. Although we cannot rule out alternative hypotheses for producing the high-Ni olivines observed in Hawaii and elsewhere, these processes or materials are unnecessary to account for NiO enrichments in olivine.