Mineral-melt calcium isotope fractionation factors constrained using ab initio molecular dynamics simulations and their implications to calcium isotope effects during partial melting in the upper mantle

Abstract

To better constrain the Ca isotope effect during partial melting of Earth’s mantle, we used ab initio molecular dynamic simulations to calculate the equilibrium mineral-silicate melt Ca isotope fractionation factors for the major Ca-bearing minerals of the upper mantle (orthopyroxene, clinopyroxene, olivine, and garnet), as well as plagioclase. We found that mineral-melt Ca isotope fractionation factors are dependent on pressure, temperature, and mineral major element compositions, but not the silicate melt composition. Specifically, our calculations show that under equilibrium, clinopyroxene has a slightly heavier Ca isotope composition compared to silicate melt, consistent with the inference of published research that studied the Ca isotope effects during basaltic magma evolution.

We then utilized the calculated mineral-melt Ca isotope fractionation factors to model the Ca isotope effects on both silicate melts and residues during partial melting of spinel peridotite at 1–2 GPa, garnet peridotite at 3–7 GPa, and garnet pyroxenite/eclogite at 2–5 GPa. Our model predicts that silicate melts only have δ^44/40Ca up to 0.12 lower than their source value, consistent with previous estimates. Importantly, partial melts of spinel peridotite, garnet peridotite, and garnet pyroxenite/eclogite exhibit overlapping δ^44/40Ca values, if their mantle sources have the same δ^44/40Ca. Partial melting alone cannot explain the full range of δ^44/40Ca values observed in natural basalts, and at least part of this variation must reflect δ^44/40Ca variation of their mantle sources and/or other processes such as crystal fractionation.

Our calculations show that melting residues always have δ^44/40Ca higher than their mantle source, with the highest δ^44/40Ca at 0.40 higher than their source value. This range is much smaller than that observed in natural ultramafic rocks that might represent melting residues. In addition, the range and direction of inter-mineral Ca isotope fractionation factors predicted in our modeled residues for the mineral pairs orthopyroxene-clinopyroxene and garnet-clinopyroxene are much more restricted than those observed in natural ultramafic rocks, including peridotites and pyroxenites/eclogites. Therefore, most natural ultramafic rocks have likely experienced more complicated petrogeneses than partial melting.