Abstract
DFT at the GGA, GGA+U and hybrid functional levels were used to investigate thousands of different Al and Fe3+ configurations of MgSiO3–FeAlO3 (MS–FA) and MgSiO3–FeAlO3–Al2O3 bridgmanite at deep mantle conditions. Comparison of the different functionals and atomic charge analysis suggests that GGA, frequently used to explain high to low spin transitions observed in several Mössbauer and X-ray emission spectroscopy experiments, is hampered by spurious self-interaction errors in the exchange-correlation energy. Configurational Boltzmann averaging shows that the B site is thermally inaccessible to Fe3+ at the GGA+U and hybrid levels, and we find no evidence for a spin-pairing transition in fully (thermodynamically) equilibrated samples of bridgmanite, even at the lowermost mantle conditions. The comparison of the cation radii of Fe3+ and Mg supports a spin transition accompanied by a site exchange, but the flexibility of FeO bonds to locally adapt promotes the incorporation of iron in the irregularly coordinated A-site. The concept of ionic radii is therefore unsuitable for analysis of spin state and site exchange in bridgmanite at these conditions. Consistent with previous computational work and experimental studies with glass and gel as starting material, we find that ferric iron kinetically trapped at the B site undergoes a spin transition under lowermost mantle conditions. In bridgmanite with mole fraction of Fe3+>Al a charge-balancing amount of low spin Fe3+ will be thermodynamically stable at the B site, but because bridgmanite in peridotitic and basaltic lithologies mostly has Al/Fetotal above unity, FA with high spin Fe3+ in the A-site will be the dominant iron component. The lack of a Fe3+ spin transition in the FA-component has important implications for bridgmanite–ferropericlase partitioning of iron and magnesium and the mineral physics of the lowermost mantle.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have