Program > Browse abstracts by speaker > Schulze Maximilian

Oxygen self-diffusion in Davemaoite studied by machine learning molecular dynamics simulations
Maximilian Schulze  1@  , Gerd Steinle-Neumann  1@  
1 : Bavarian Research Institute of Experimental Geochemistry and Geophysics  (Bayerisches Geoinstitut)

Davemaoite (CaSiO3 perovskite) is a major constituent of the Earth's lower
mantle. Its transport properties are therefore crucial for a comprehensive un-
derstanding of deep Earth processes. However, experimental studies of the
properties of Davemaoite are severely limited by the inability to preserve its
crystal structure during quenching. This concerns in particular the investiga-
tion of element diffusivities, which are typically determined by ex situ analysis
of samples recovered from experiments conducted at high pressure (P ) and tem-
perature (T ).
A compelling alternative to experimental methods for studying chemical dif-
fusion in Earth materials is presented by ab initio molecular dynamics (AIMD)
simulations. Nevertheless, the considerable computational resources necessary
for these simulations often restrict their application to relatively small system
sizes and short timescales. This restriction frequently precludes a comprehen-
sive evaluation of finite size effects and the statistical significance of the results.
To overcome these limitations, we employ a machine learning potential trained
on high-quality density functional theory data, allowing us to perform MD sim-
ulations on extended time and length scales while maintaining quantum-level
accuracy. This approach is utilized to examine oxygen self-diffusion in Dave-
maoite across the P -T range of the lower mantle. The results are then discussed
in the context of the timescales of redox equilibria in subducted slabs, thereby
providing a novel perspective on the dynamics of the deep Earth.


Loading... Loading...