dataset in Li et al: Calcium isotope effects during partial melting of the mantle
Description
This file contains the tables presenting data in Li et al: Calcium isotope effects during partial melting of the mantle. To better constrain the Ca isotope fractionation during partial melting of Earth’s mantle, we used ab initio molecular dynamic simulations to calculate the equilibrium mineral-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 melt composition. Specifically, our calculations show that under equilibrium, clinopyroxene has slightly heavier Ca isotope composition compared to melt, consistent with the inference of published research that studied the Ca isotope effect during magma evolution. We then utilized the calculated mineral-melt Ca isotope fractionation factors to model the Ca isotope composition evolutions of both melts and residues during partial melting of spinel peridotite at 1-2 GPa, garnet peridotite at 3-7 GPa, and garnet pyroxenite/eclogite at 2.5-20.7 GPa. Our model predicts that, silicate melts only have δ44/40Ca up to 0.11‰ lower than their source value, consistent with previous estimates. Importantly, patrial melts of spinel peridotite, garnet peridotite, and garnet pyroxenite/eclogite exhibit overlapping δ44/40Ca values, if their mantle sources have the same δ44/40Ca. In contrast to silicate melts, we predict that carbonatitic melts have δ44/40Ca up to 0.35‰ lower than their mantle source, and the largest δ44/40Ca effect is found in carbonatitic melts produced under mantle transition zone conditions. In both cases, partial melting alone cannot explain the full range of δ44/40Ca observed in natural basalts and carbonatites, and at least part of this variation must reflect δ44/40Ca variation in their mantle sources. Our calculations show that melting residues always have δ44/40Ca higher than their mantle source, with the highest δ44/40Ca at 0.40‰ higher than their source value. This range is much smaller than that observed from 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 petrogenesis than partial melting.
Files
Steps to reproduce
Ab initio molecular dynamic (AIMD) simulations were performed on the nine melts (Table 1) using Vienna Ab-initio Simulation Package (VASP) (Kresse et al., 1994; Kresse et al., 1996) to obtain a series of trajectory configurations that were used to calculate the properties of the melts, such as density, bonding, phonons, and Ca isotope β value. In all our calculations, we adopted the projector-augmented wave (PAW) method (Blöchl, 1994; Kresse et al., 1999) and PBE parameterized generalized-gradient approximation (GGA) (Perdew et al., 1996) pseudopotentials (GGA-PBE). The energy cutoff was 600 eV. Canonical ensemble (NVT) and Nose-Hoover thermostat (Nosé, 1984; Hoover, 1985) were used to perform the AIMD simulations of the nine melts at 3000 K. The simulation time lasts beyond 80 ps with a time step of 1 fs. The NVT ensemble was used and the volumes of nine melts (Table 1) were assigned so that their MD simulation pressures (P_md) were close to ~3 GPa. Besides, the BA, DA, TH, TE and YA melts were also simulated under variable volumes to evaluate the pressure effects on basaltic and carbonatite melts (Table S1). To obtain the Ca isotope β value of a melt, the “snapshot” method was used. In this method, a number of instantaneous configurations (snapshots) were extracted from the AIMD trajectories of a melt. Each configuration was ionically relaxed under static conditions to calculate its phonon density of states. Finally, the Ca isotope β value was calculated from the vibrational frequencies using the Urey equation (Bigeleisen et al., 1947).