Sur ce site

Sur le Web du CNRS


Accueil du site > Productions scientifiques > Séminaires à PHENIX > 2010 > Séminaire Jeune Chercheur 15.12.2010 - 10:00

Séminaire Jeune Chercheur 15.12.2010 - 10:00

par Guillaume Mériguet - 10 décembre 2010

Volker Haigis (équipe Molecular Modeling of Geomaterials, GFZ Potsdam) présentera un séminaire le mercredi 15/12/2010 à 10h00 dans la bibliothèque du PECSA (7e étage, batiment F, porte 754) intitulé :

Atomistic simulation approach to trace element partitioning between silicate melts

Résumé :

Element partitioning, i.e. the preferred incorporation of certain cations into a given chemical environment rather than into other coexisting phases, plays a fundamental role in processes such as crystallization of silicate magmas or planetary differentiation. The lattice strain model proposed by Blundy & Wood (1994) attempts to explain the observed partitioning of trace elements between coexisting minerals and melts entirely in terms of the strain induced in the host crystal lattice by the size misfit of the incorporated cation : the closer the cation’s ionic radius to the ideal ionic radius in the host lattice, the more easily it enters the mineral, otherwise it is enriched in the melt. The influence of the melt structure on the partitioning, on the other hand, is neglected by the lattice strain model.

However, recent experiments by Prowatke & Klemme (2005) challenge this view : they show that, for a given silicate crystal structure, the partitioning of many trace element cations (e.g. Y) depends strongly on the composition of a coexisting silicate melt, in particular on the ratio of the Al to alkali and alkaline earth contents. It was suggested that this ratio influences the framework structure of the melt and hence also the local environment of incorporated ions, which in turn determines the thermodynamics of partitioning. A detailed understanding of atomic-scale structures in melts leading to the observed behavior is still missing.

Here, we present a combined DFT and classical potential approach to trace element partitioning in silicate melts, taking Y as a trace element. First, a classical ionic interaction potential, including ionic polarization terms, is established for the system Y2O3 - CaO - Al2O3 - SiO2. Its parameters are obtained by fitting ionic forces, dipoles and stresses to those calculated with DFT. The resulting potential is tested by comparing its predictions for neutron structure factors and radial distribution functions of oxide melts to experimental data and DFT calculations.

Then the potential is used in atomistic simulations to investigate the coordination environment of Y in silicate melts of different compositions. As compositional variable the ratio Al2O3/CaO is chosen, which is expected to drive the observed variation of the partition coefficient. We explore the relation between melt composition and the local melt structure around Y, and discuss qualitatively its relation to the partitioning behavior of Y. This constitutes the initial step towards a quantitative atomic-scale understanding of trace element partitioning.

References :

Blundy, J. & Wood, B. (1994), Prediction of crystal-melt partition coefficients from elastic moduli, Nature 372, 452

Prowatke, S. & Klemme, S. (2005), Effect of melt composition on the partitioning of trace elements between titanite and silicate melt, Geochim. Cosmochim. Acta 69, 695