期刊
EARTH AND PLANETARY SCIENCE LETTERS
卷 273, 期 1-2, 页码 38-47出版社
ELSEVIER
DOI: 10.1016/j.epsl.2008.06.005
关键词
post-magnesite; density functional theory; ab initio simulations; crystal structure prediction; evolutionary algorithm; high pressure
资金
- Supercomputer Centre of Russian Academy of Sciences
- Skif Cyberia supercomputer of Tomsk State University (Russia)
- Swiss National Science Foundation [200021-111847/1]
- ETH Zurich Research Equipment Programme
- KEK [2005G122]
- Spring-8 [2006A1412]
- Spanish Ministry of Education [MAT2005-05216, FIS2006-12117-CO4-01]
Most of the oxidized carbon in the Earth's lower mantle is believed to be stored in the high-pressure forms of MgCO3 and/or CaCO3 or possibly even CO2. Recently, through ab initio evolutionary simulations and high-pressure experiments. a complete picture of phase transformations of CaCO3 at mantle pressures was obtained. Here, using the same approach, we investigate the high-pressure structures of MgCO3. Two new structure types were predicted to be stable in the relevant pressure range: one at 82-138 GPa and the other above 138 GPa. Both phases contain rings of corner-sharing CO4-tetrahedra. These predictions were largely confirmed by the experiments presented here. A number of structurally very different, but energetically competitive metastable polymorphs were found and reveal complex high-pressure chemistry of MgCO3, in contrast to CaCO3. For CO2, from 19 GPa to at least 150 GPa, we find beta-cristobalite structure to be stable. Differences between high-pressure tetrahedral carbonates and low-pressure silicates are discussed in terms of rigidity of the T-O-T angles (flexible when T=Si and stiff when T=C). We show that through most of the P-T conclitions of the mantle, MgCO3 is the major host of oxidized carbon in the Earth. We discuss the possibility of CO2 release at the very bottom of the mantle, which could enhance partial melting of rocks and explain the geodynamical differences between the Earth and Venus. (C) 2008 Elsevier B.V. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据