Cation Ordering in Fe-bearing Silicate Perovskite (Bridgmanite) and its Role in Disproportionation

Tuesday, 16 December 2014: 4:15 PM
Aaron S Wolf1, Razvan Caracas2, Paul D Asimow3 and Jean-Philippe Harvey3, (1)University of Michigan, Ann Arbor, MI, United States, (2)Ecole Normale Supérieure Lyon, Lyon, France, (3)California Institute of Technology, Pasadena, CA, United States
Previous theoretical studies have investigated the composition-dependence of the spin transition in Fe-bearing MgSiO3 perovskite, finding that the transition pressure is extremely sensitive to atomic configuration. Computational constraints limited these studies to supercell structures containing only one to a few iron atoms in a handful of configurations. Bengston (2008) handled this issue using special quasirandom structures, which approximate a randomized macroscopic sample. Umemoto (2008) considered a set of configurations in large supercells, finding that iron in perovskite prefers to cluster. Although these studies showed that the energetic differences between configurations can be as large as ~10 mHa (3000K thermal energy equivalent), none have yet attempted to fold the configuration-dependence into a thermodynamic ordering model. In this presentation, we report an adaptation of the cluster-site-approximation (Oates, 1999) that enables efficient calculation of order-dependent phase diagrams for perovskite, given a reasonable set of density functional theory calculations of modest supercell sizes.

Recently, Zhang (2014) reported the occurrence of perovskite disproportionation in high P-T diamond anvil cell experiments. The studies discussed above found that intermediate iron compositions were unstable relative to endmembers, implying potential exsolution, but no experimental evidence for it existed at the time. Zhang (2014) showed that all Fe-bearing perovskite samples with 10% to 40% Fe/[Fe+Mg] underwent exsolution of a hexagonal iron-rich phase when the sample was compressed to 100 GPa prior to heating. This poses a significant apparent contradiction: most studies have found using progressive annealing that iron-bearing perovskite is stable up to the post-perovskite transition (e.g. Dorfman 2013), whereas Zhang (2014) showed that the wait-and-heat method consistently induces disproportionation. We discuss how this is a likely outcome of the ordering tendencies of perovskite and show how the development of iron-rich clusters tends to stabilize perovskite against disproportionation. Finally, we compare the different roles ordering might play under experimental conditions and deep within the Earth.