A large amount of research has been done in which the magnetocrystalline anisotropy energy for fcc Ni and bcc Fe was calculated based on the electronic structure of these elements. Unfortunately; the results of these studies don't agree with each other and also differ from the experimental observation. In a previous thesis the effects of numerical errors in the Brillouin zone integrations were investigated. The results of that work explain why different calculations give different results, but do not explain the difference with experiment. The conclusion was that the underlying bandstructure, which was calculated using standard approximations, was not correct. The bandstructure of these elements will be different when improved prescriptions for the exchange-correlation energy are used. There is, however, no clear indication along which lines this approximation should be improved. Here we have taken a different approach to change the bandstructure. We suspected that some important interactions between different atomic orbitals are either ignored or miscounted. In this work, we examined the sensitivity of the energy on the interaction between those orbitals and studied in detail the consequences of changes in some interaction parameters which gave rise to a large energy change. The main result of this work is a better understanding of the relation between changes in the electronic structure in k-space and the resultant change in the magnetocrystalline anisotropy energy. In addition, this work takes another step in trying to find a better understanding how the magnetocrystalline anisotropy energy relates to interactions between neighboring atoms. / Graduation date: 2000

No description available.