Dynamical Mean Field Theory

Density-Functional theory, while being immensely popular thanks to its simplicity, nevertheless is limited in its reliability. The QuasiParticle Self-Consistent GW approximation, while more demanding than DFT, is vastly more reliable than DFT, or GW theory based on DFT, for calculation of optical properties in weakly correlated systems.

In this soon to appear paper in Phys Rev B Rapid Communications, the validity of QSGW for two magnetic transition metals are critically examined : Fe and Ni. We show spectacular agreement with a variety of experiments for states near the Fermi level in Fe [Figure fe-bands-with-arpes.eps]; the theory is good enough that we can identify limits to the standard interpretation of Angle Resolved Photoemission measurements. For Ni, agreement is less good because of spin fluctuations missing in both GW and Density Functional theory. By constructing a novel form of QSGW+Dynamical Mean Field theory, we can include spin fluctuations with a minimum of ambiguity. Very good agreement with Angle Resolved Photoemission measurements is found, and a satisfactory description of the role of spin fluctuations in modifying the magnetic moment obtained.

The Figure [bnds-mom-ni.eps] compares the energy band structure of Ni in QSGW (solid lines) and LDA (dotted) and ARPES data. Red arrows highlight the discrepancy in the exchange splitting ∆Ex at near L and X. (right) QSGW+DMFT bands with method S (solid) and QSGW+Beff (dashed). (inset) ∆Ex at L as a function of M obtained by adding an external magnetic field to the QSGW or LSDA potential (see text).