The quasiparticle band structures of 3d transition metals, ferromagnetic Fe, Ni and paramagnetic Cu, are calculated by the GW approximation. The width of occupied 3d valence band, which is overestimated in the LSDA, is in good agreement with experimental observation. However the exchange splitting and satellite in spectra are not reproduced and it is required to go beyond the GW approximation. The effects of static screening and dynamical correlation are discussed in detail in comparison with the results of the static COHSEX approximation. The dynamical screening effects are important for band width narrowing.
Download