Non-empirical prediction of the photophysical and magnetic properties of systems with open d- and f-shells based on combined Ligand Field and Density Functional Theory (LFDFT)


Fanica Cimpoesu, Harry Ramanantoanina, Benjamin Herdem, Werner Urland and Claude Daul


Computational Chemistry Lab, University of Fribourg / Switzerland



Abstract:

Despite the important growth of ab initio and computational technics, ligand field theory in molecular science or crystal field theory in condensed matter offers the most intuitive way to calculate multiplet energy levels arising from systems with open shells d and/or f electrons. We developed this last decade a ligand field treatment of inorganic molecular modelling taking advantages of the dominant localisation of the frontier orbitals within the metal-sphere. This feature, being observed in any inorganic coordination compounds especially if treated by Density Functional Theory calculation, allows the determination of the electronic structure and properties with a surprising good accuracy. In ligand field theory, the theoretical concepts consider only a single atom centre; and treat its interaction with the chemical environment essentially as a perturbation. Therefore success in the simple ligand field theory is no longer questionable, while the more accurate molecular orbital theory does in general over-estimate the metal-ligand covalence, thus yields wave functions that are too delocalized. Although LF theory has always been popular as a semi-empirical method when dealing with molecules of high symmetry e.g. cubic symmetry where the number of parameters needed is reasonably small (3 or 5), this is no more the case for molecules without symmetry and involving both an open d- and f-shell (number of parameters is roughly 90). However, the combination of LF theory and Density Functional (DF) theory that we introduced twenty years ago can easily deal with complex molecules of any symmetry with two and more open shells. The accuracy of these predictions from 1st principles achieves quite a high accuracy (less than 5 percent) in terms of states energies. Hence, this approach is well suited to predict the magnetic and photo-physical properties arbitrary molecules and materials prior to their synthesis, which is the ultimate goal of each computational chemist. We will illustrate the performance of LFDFT for the design of phosphors that produces light similar to our sun and predict the magnetic anisotropy energy of single ion magnets.