Density‐Pressure Profiles of Fe‐Bearing MgSiO₃ Liquid: Effects of Valence and Spin States, and Implications for the Chemical Evolution of the Lower Mantle

Abstract

Density is a key property controlling the chemical state of Earth's interior. Our knowledge about the density of relevant melt compositions is currently poor at deep-mantle conditions. Here we report results from first-principles molecular-dynamics simulations of Fe-bearing MgSiO 3 liquids considering different valence and spin states of iron over the whole mantle pressure conditions. Our simulations predict the high-spin to low-spin transition in both ferrous and ferric iron in the silicate liquid to occur gradually at pressures around 100 GPa. The calculated iron-induced changes in the melt density (about 8% increase for 25% iron content) are primarily due to the difference in atomic mass between Mg and Fe, with smaller contributions (<2%) from the valence and spin states. A comparison of the predicted density of mixtures of (Mg,Fe)(Si,Fe)O 3 and (Mg,Fe)O liquids with the mantle density indicates that the density contrast between the melt and residual-solid depends strongly on pressure (depth): in the shallow lower mantle (depths < 1,000 km), the melt is lighter than the solids, whereas in the deep lower mantle (e.g., the D″ layer), the melt density exceeds the mantle density when iron content is relatively high and/or melt is enriched with Fe-rich ferropericlase.How much iron changes the melt density also depends on its valence and spin states as they can influence the volume to different extents. Iron in bridgmanite (Mg,Fe)(Si,Fe)O 3 exists in ferrous and ferric oxidation KARKI ET AL. 3959Key Points:• First-principles simulations predict pressure-induced spin transition of iron in silicate melt with modest effect on melt density • Silicate melt density increases strongly with iron content, more so at higher pressures • While the melt density is low in shallow lower mantle, dense melts could be buoyantly stable at deep mantle (D″ layer) Supporting Information:• Supporting Information S1

Fate of Hydrous Fe‐Rich Silicate Melt in Earth's Deep Mantle

“…So the density crossover is essentially dominated by Fe enrichment in the melt. These evolution scenarios are similar to the previous studies for a water‐free magma ocean (e.g., Caracas et al, ; Karki et al, ; Miyazaki & Korenaga, ) when φ > 10%. During such early stages of crystallization, the details of how crystal settled are still debated and model dependent (e.g., Abe, ; Solomatov & Stevenson, ).…”

“…The solid curves represent the density profiles from the EoS fit and the ideal water‐melt solution model (equation ). The dashed curve is the density derived from Karki et al (). (b) Density contrast between hydrous and anhydrous E melts as a function of the water content at 4,000 K. The present results for E melt (solid circles) show a linear trend (solid line).…”

“…As shown in Figure a, the density profiles of hydrous E melts are systematically shifted downward from the anhydrous melt density as water lowers the melt density considerably. We can also evaluate the density of anhydrous E melt by combining the results for bridgmanite and ferropericlase liquids as done by Karki, Ghosh, Maharjan, et al (). Thus derived density profile almost matches with the one directly obtained in this study (Figure a).…”

“…Given that both K Fe and D H2O values are much smaller than 1, melt should be much more enriched in both Fe and water than the solid at equilibrium. To demonstrate the importance of compositional effects, we use equation to compute the melt density as a function of water content and Fe content (Fe # ) at the CMB condition, namely, 135 GPa and 4,000 K. We also use the density model from Karki et al () to calculate ρ 0 for different Fe contents with the low‐ to high‐spin Fe ratio set at 80:20. Figure shows a contour of density difference equal to zero between a silicate melt and solid (i.e., neutral density buoyancy) at the CMB condition.…”

“…Following Karki et al (2018), we consider a scenario for crystallization of a fully molten Earth, assuming equilibrium crystallization (Solomatov & Stevenson, 1993b). We further assume initial uniform iron and water contents in the magma ocean as X Fe = 0.1 and X H2O = 200 ppm, respectively.…”