First-principles calculations for a model Hamiltonian treatment of hybridizing light actinide compounds
Abstract
A procedure, previously applied to Ce(f1) systems, for calculating the components of an Anderson-model lattice Hamiltonian has been extended to hybridizing actinide (fn>1) systems and applied to some hybridizing Pu compounds with magnetic behavior of great current interest. There are some differences in the treatment of actinide systems compared with cerium systems, arising both from the greater spatial extent of 5f states compared with 4f states and from the many-electron character of the actinide correlated quasi-ionic states. The band-f hybridization matrix elements and band energies in the Anderson Hamiltonian are obtained from warped-muffin-tin linear muffin-tin-orbital band-structure calculations with Pu 5f states treated as ``just bound'' corelike states. The calculations are performed for PuAs, PuSb, PuBi, and PuTe. Pu 5f energy levels, relative to the Fermi energy, are calculated for PuSb from total-energy supercell calculations with four, five, and six electrons in the Pu 5f core. We analyze the source in the electronic structure of trends between materials in hybridization-induced effects by considering the behavior of two quantities: (1) f-state resonance widths, characterizing the strength of hybridization, and (2) the density of states at the Fermi energy, characterizing the number and character of band states available for hybridization. Besides assessing the trend in hybridization effects in relationship to changes in the electronic structure between materials, the results are used to evaluate two quantities arising in the model Hamiltonian treatment from hybridization between Pu f states and band states: a shift in the crystal-field levels of the Pu(f5,J=5/2) multiplet and a two-ion exchange interaction between Pu ions on different sites. We compare the calculated quantities with experimental results and with previous phenomenological calculations of the magnetic behavior of PuSb, and discuss the implications of our calculations for the magnetic behavior to be expected in these materials. Experimentally, the magnetic behavior of PuSb and PuTe are quite different; so it is interesting that changes in the calculated magnitude and range dependence of the exchange parameters between PuSb and PuTe indicate a qualitative change in the nature of the expected magnetic behavior.
- Publication:
-
Physical Review B
- Pub Date:
- September 1990
- DOI:
- 10.1103/PhysRevB.42.4682
- Bibcode:
- 1990PhRvB..42.4682W
- Keywords:
-
- 71.28.+d;
- 75.10.Dg;
- 75.30.Et;
- 75.30.Mb;
- Narrow-band systems;
- intermediate-valence solids;
- Crystal-field theory and spin Hamiltonians;
- Exchange and superexchange interactions;
- Valence fluctuation Kondo lattice and heavy-fermion phenomena