Comparison of results from parametrized configuration interaction (PCI-80) and from hybrid density functional theory with experiments for first row transition metal compounds

Different methods and schemes have been tested for the difficult class of first row transition metal complexes. The systems investigated are the M(+), MH(+), MCH(3)(+), and MCH(2)(+) systems for the entire row and the Ni(CO)(x) systems with x = 1-4. In general quite satisfactory results are obtained both at the PCI-80 and hybrid density functional levels. In particular, for the MH(+) and MCH(3)(+) systems the PCI-80 average deviation to experiments is of the same size as the uncertainty in the experiments. The MCH(2)(+) systems are somewhat more difficult to describe and a rather large error is found for chromium at the PCI-80 level, due to a large coupling of exchange energy loss and change of correlation energy resulting from the formation of two covalent d-bonds. Scaling at the modified coupled pair functional (MCPF) level is also compared to scaling at the coupled cluster singles and doubles (CCSD) level. In most cases very similar results are obtained, but classes of systems can be identified where scaling works better at the MCPF level and other classes where it works better at the CCSD level. (C) 1996 American Institute of Physics.