No Arabic abstract
We propose a new ab initio method designed for the accurate calculation of effective exchange integrals between atoms with numerous open-shells. This method applies to ferromagnetic as well as antiferromagnetic exchange, direct or ligand-mediated exchange. Test calculations on high spin transition metal oxides such as KNiF3, Ba2CoS3 or YMnO3 exhibit a very good accuracy compared either to the best ab initio calculations --when those are feasible-- and with experimental evaluations.
The present paper proposes the direct calculation of the microscopic contributions to the magneto-electric coupling, using ab initio methods. The electrostrictive and the Dzyaloshinskii-Moriya contributions were evaluated individually. For this purpose a specific method was designed, combining DFT calculations and embedded fragments, explicitely correlated, quantum chemical calculations. This method allowed us to calculate the evolution of the magnetic couplings as a function of an applied electric field. We found that in $rm YMnO_3$ the Dzyaloshinskii-Moriya contribution to the magneto-electric effect is three orders of magnitude weaker than the electrostrictive contribution. Strictive effects are thus dominant in the magnetic exchange evolution under an applied electric field, and by extension on the magneto-electric effect. These effects remain however quite small and the modifications of the magnetic excitations under an applied electric field will be difficult to observe experimentally. Another important conclusion is that the amplitude of the magneto-electric effect is very small. Indeed, it can be shown that the linear magneto-electric tensor is null due to the inter-layer symmetry operations.
We used quantum chemical ab initio methods to determine the effective parameters of Hubbard and $t-J$ models for the $rm Na_{x}CoO_2$ compounds (x=0 and 0.5). As for the superconducting compound we found the $a_{1g}$ cobalt orbitals above the $e_g^prime$ ones by a few hundreds of meV due to the $e_g^prime$--$e_g$ hybridization of the cobalt $3d$ orbitals. The correlation strength was found to increase with the sodium content $x$ while the in-plane AFM coupling decreases. The less correlated system was found to be the pure $CoO_2$, however it is still strongly correlated and very close to the Mott transition. Indeed we found $U/tsim 15$, which is the critical value for the Mott transition in a triangular lattice. Finally, one finds the magnetic exchanges in the $rm CoO_2$ layers, strongly dependant of the weak local structural distortions.
We propose an approach for the ab initio calculation of materials with strong electronic correlations which is based on all local (fully irreducible) vertex corrections beyond the bare Coulomb interaction. It includes the so-called GW and dynamical mean field theory and important non-local correlations beyond, with a computational effort estimated to be still manageable.
RESPACK is a first-principles calculation software for evaluating the interaction parameters of materials and is able to calculate maximally localized Wannier functions, response functions based on the random phase approximation and related optical properties, and frequency-dependent electronic interaction parameters. RESPACK receives its input data from a band-calculation code using norm-conserving pseudopotentials with plane-wave basis sets. Automatic generation scripts that convert the band-structure results to the RESPACK inputs are prepared for xTAPP and Quantum ESPRESSO. An input file for specifying the RESPACK calculation conditions is designed pursuing simplicity and is given in the Fortran namelist format. RESPACK supports hybrid parallelization using OpenMP and MPI and can treat large systems including a few hundred atoms in the calculation cell.
The magnetic properties of the transition metal monoxides MnO and NiO are investigated at equilibrium and under pressure via several advanced first-principles methods coupled with Heisenberg Hamiltonian MonteCarlo. The comparative first-principles analysis involves two promising beyond-local density functionals approaches, namely the hybrid density functional theory and the recently developed variational pseudo-self-interaction correction method, implemented with both plane-wave and atomic-orbital basis sets. The advanced functionals deliver a very satisfying rendition, curing the main drawbacks of the local functionals and improving over many other previous theoretical predictions. Furthermore, and most importantly, they convincingly demonstrate a degree of internal consistency, despite differences emerging due to methodological details (e.g. plane waves vs. atomic orbitals)