Chemical Bonding, 10Dq Parameter and Superexchange in the Model Compound KNiF 3 .

The cubic field splitting parameter, 10Dq, plays a central role in the ligand field theory on insulating transition metal compounds. Experimental data obtained in the last 50 years prove that 10Dq is highly dependent on changes of the metal-ligand distance, R, induced by chemical or applied pressures. Despite this fact has important consequences on optical and magnetic properties of such compounds, its actual origin is still controversial. Seeking to clarify that crucial issue, this work is focused on KNiF 3 , a reference system among insulating transition metal compounds. By means of first principles calculations we show that, contrary to what is usually thought, the R-dependence of 10Dq arises neither from the crystal field contribution nor from the covalent admixture of 3d(Ni) with valence 2p(F) orbitals. Indeed, we prove that it is mainly due to the residual covalency with deep 2s(F) orbitals, highly sensitive to R variations. As a salient feature the present calculations show that the 3d-2pσ and 3d-2pπ admixtures raise practically equal the energy of antibonding e g and t 2g orbitals of NiF 6 4- units in KNiF 3 thus leading to a null contribution to 10Dq. This conclusion is not significantly altered when considering the change of covalency on passing from the ground state 3 A 2 (t 2g 6 e g 2 ) to the excited state 3 T 2 (t 2g 5 e g 3 ). The different influence of chemical bonding on the superexchange constant, J, and 10Dq is also discussed in a second step. It is pointed out that the strong dependence of J upon R can hardly be explained through the behavior of the 3d-2pσ covalency derived for a single NiF 6 4- unit. For the sake of clarity, the meaning of 10Dq is also briefly analyzed.