Diffusion and thermodynamics of defects in alkali feldspar
simulated through machine learning force fields
Alexander Gorfer
Faculty of Physics, University of Vienna, Austria &
Department of Lithospheric Research, University of Vienna, Austria
Exsolution microstructures of alkali feldspar (Na,K)AlSi3O8 which form during slow cooling
of high-temperature rocks encode the rocks thermal history, but we do not yet fully know
how to interpret them quantitatively. In this talk I will present computer simulations driven
by machine learning force fields that shed light on the microscopic mechanism of diffusion
that ultimately controls the evolution of the exsolution microstructures.
In Na-feldspar we find a new kind of Na+-Na+ dumbbell interstitial point defect and predict
that this is the most stable interstitial defect over all temperatures. Our simulations of the
dynamics of this defect combined with the calculated defect concentration quantitatively
agree with experimental sodium self-diffusion data, which allows us to rectify past investigations
on the mechanism and reconcile prior works on the anisotropy of cation diffusion.
I will also give a short overview of our recent work in which we use a combination of molecular
dynamics and Monte Carlo simulations to calculate the mixing thermodynamics of Nafeldspar
and K-feldspar above the alkali feldspar solvus. Our model correctly aligns with the
literature data in that increasing Al-Si disorder corresponds to a more ideal mixture and we
predict that the degree of Al-Si disorder is less relevant as long as disorder persists.
Mots clés : alkali feldspar defects diffusion machine learning force fields
Informations
- Patrick Cordier (patrick.cordier)
-
- 25 octobre 2024 13:26
- Webinaire
- Anglais