Japan Geoscience Union Meeting 2023

Presentation information

[E] Online Poster

S (Solid Earth Sciences ) » S-IT Science of the Earth's Interior & Techtonophysics

[S-IT16] Deep Earth Sciences

Fri. May 26, 2023 9:00 AM - 10:30 AM Online Poster Zoom Room (2) (Online Poster)

convener:Jun Tsuchiya(Geodynamics Research Center, Ehime University), Kenji Ohta(Department of Earth and Planetary Sciences, Tokyo Institute of Technology), Kenji Kawai(Department of Earth and Planetary Science, School of Science, University of Tokyo), Tsuyoshi Iizuka(University of Tokyo)

On-site poster schedule(2023/5/25 17:15-18:45)

9:00 AM - 10:30 AM

[SIT16-P19] Thermal conductivity of hydrous Wadsleyite determined by non-equilibrium molecular dynamics based on machine learning

*Dong Wang1, Zhongqing Wu1, Xin Deng1 (1.University of Science and Technology of China)

Keywords:Thermal conductivity, Hydrous Wadsleyite, Machine learning

Subduction slabs are heated primarily by thermal conduction. The thermal conductivity of minerals in slabs influences many subduction processes by controlling the temperature of slabs. Recent studies have demonstrated that water significantly reduces the thermal conductivity of olivine and ringwoodite. However, the effect of water on the thermal conductivity of wadsleyite, which can contain significant amounts of water in the subduction slabs, remains unknown. Measurements of the thermal conductivity of minerals at high temperature and pressure are still challenging. Molecular dynamics simulation has been widely used to determine the thermal conductivity in various systems with the advantage that anharmonicity is fully included. However, first-principles molecular dynamics lead to large finite-size effects due to the small number of atoms that can be studied. Machine learning potential trained from first-principles data combines the accuracy of first-principles with the computational efficiency of classical potentials, providing an alternative and promising way to simulate large systems and calculate thermal conductivity.
In this study, we developed a high-quality machine learning potential for wadsleyite with data from first-principles calculations. Combining non-equilibrium molecular dynamics simulations and machine learning potential, we predicted the thermal conductivity of hydrous and dry wadsleyite at high temperature and pressure. We found that the thermal conductivity of wadsleyite is anisotropic and is reduced by ~10% in the pressure and temperature conditions of the MTZ by the presence of 0.81wt.% water. Heat flow to slabs may follow the direction with relatively low thermal conductivity due to the anisotropy and lattice-preferred orientation of olivine and wadsleyite, which further prevents the heating of slabs. Low temperatures increase the survival depth of hydrous and metastable minerals, improving the ability of slabs to transport water, and increasing the maximum depth of deep-focus earthquakes.