AB-INITIO QUANTUM MECHANICAL CALCULATIONS OF POLARISED NEUTRON DIFFRACTION: RESULTS IN TRANSI-TION METAL COMPLEXES. P.A. Reynolds, Research School of Chemistry, Australian National University, Canberra, ACT 0200, Australia

Metal-ligand bonding in metal complexes is poorly understood. Spin density is a fundamental property, it is much modified by bonding effects, and it has important chemical consequences. Consider the instability of triplet H₂ and the stability of singlet H₂. In certain magnetically simple cases in paramagnetic crystals we can estimate many Fourier components of the spin-density, with an accuracy of ca. 1%, by use of polarised neutron diffraction. Covalence, resulting in delocalisation of positive metal-centred spin density onto ligand atoms is particularly marked in our recent results on [As(C₆H₅)₄][TcNCl₄] in which 26% of an ionic, technetium centred, 4d_xy¹ spin density has been delocalised onto the chlorines' in-plane 3p_[[pi]] orbitals. Spin-polarisation, due to electron-electron correlation, is also marked in the TcN bond. The nitrogen spin population is negative, -0.18 e. Spin-orbit coupling besides causing canting of the magnetisation, as in Cr(II) Tutton salt, can also cause large changes in the radial distribution of the spin density around the metal site, as in CoCl₄^2-.

Calculations must be able to duplicate these covalence, spin polarisation, and spin-orbit coupling effects. This requires a correlated relativistic wavefunction (example CoCl₄^2-). In some cases approximations such as the Unconstrained Hartree-Fock method may be appropriate in description of metal-ligand bonding, but other approximations such as the Restricted Hartree-Fock method never are (example TcNCl₄^-).