Electronic-structure calculations of elemental praseodymium are presented. Several approximations are used to describe the Prf electrons. It is found that the low-pressure, trivalent phase is well described using either the self-interaction corrected (SIG) local-spin-density (LSD) approximation or the generalized-gradient approximation (GGA) with spin and orbital polarization (OF). In the SIC-LSD approach the Prf electrons are treated explicitly as localized with a localization energy given by the self-interaction of the f orbital. In the GGA+ OP scheme the f-electron localization is described by the onset of spin and orbital polarization, the energetics of which is described by spin-moment formation energy and a term proportional:to the total orbital moment, L-z(2). The high-pressure phase is well described with the f electrons treated as band electrons, in either the LSD or the GGA approximations, of which the latter describes more accurately the experimental equation of state. The calculated pressure of the transition from localized to delocalized behavior is 280 kbar in the SIC-LSD approximation and 156 kbar in the GGA+OP approach, both comparing favorably with the experimentally observed transition pressure of 210 kbar.

Electronic structure calculations, based on density functional theory, demonstrate that the surface of α-Ce is spin and orbitally polarized and very similar to that obtained from a calculation for bulk γ-Ce. In contrast, the lower lying layers do not deviate from characteristics associated with bulk α-Ce. These results imply that the surface of α-Ce is γ-like. This is consistent with photoemission experiments and the calculations provide a theoretical complement to the these data. In addition to confirming the picture given by the photoemission data, we demonstrate that it is only the topmost layer of Ce atoms which have electronic properties which deviate from the bulk. This in turn shows that the chemical bonding of the surface atoms is very different compared to the bulk atoms. Our finding that it is only the topmost layer of α-Ce which deviates from bulk behaviour can probably be verified experimentally.

We have shown theoretically, by means of a full-potential LMTO method in a slab geometry, that the topmost surface layer of α-Ce is γ-like. The calculations are fully relativistic and include all electrons. Also an orbital correction to the f-electron states is included to allow for f localization. For the topmost layer the spin moment is found to be 0.70 μB for the f electrons and 0.08 μB for the d-electron states while the orbital moment from the f electrons is found to be −1.07 μB. These moments correspond well to the moments calculated for bulk γ-Ce which are 0.63, 0.07, and −0.87 μB, respectively. The subsurface and lower lying layers are calculated to have spin and orbital moments close to zero. We therefore conclude that the topmost layer on the α-Ce surface is γ-like while the subsurface and lower lying layers are similar to bulk α-Ce. Our findings explain recent photoemission experiments on this material and provide a theoretical foundation for the interpretation of these experiments. The fact that our theory assumes that α-Ce has delocalized f electrons, and the agreement with the interpretation of photoemission experiments, gives support that the Mott transition model for the α→γ phase transition is correct.

We predict an inwards surface relaxation of the (101̄0) surface of Be. The surface layer relaxes inwards with 25% of the inter planar distance, a rather large value. However, the corresponding change of nearest-neighbour interatomic distances is smaller, of the order 2–6%. We show that the relaxation is correlated with a pronounced surface state in the vicinity of the Fermi level. By following the inward relaxation of the surface layer, we show that the total energy is lowered mainly due to the one-particle energy. The most preferable termination of the Be(101̄0) surface is shown to be the so-called A termination. Also, the termination of the Be(0001) is investigated and we conclude that the hcp termination is preferable compared to the fcc termination.

We predict a giant surface relaxation of the (1010) surface of Be. The surface layer relaxes inwards with 25%, which we show is due to a pronounced surface state close to the Fermi level. We also predict that the (1010) surface at Be is terminated such that the surface has 8 nearest neighbours (A termination).

The calculated electronic structure and charge density of UC and US are reported. We also report calculations on the elastic constants of the cubic compounds UC, US, and UTe using the local-spin-density approximation (LSDA) to the exchange and correlation potential. Good agreement between calculations and experiment has been obtained for UC and US suggesting that the chemical bonding is well reproduced by LSDA for these two compounds. In contrast, for UTe the calculations do not reproduce the measured negative Poissons ratio (negative C-12). We suggest that the failure to describe the elasticity of UTe reflects the weakness of LSDA in describing accurately electron systems that are strongly correlated and we speculate that UTe is an anomalous, possibly mixed valent, system.

We demonstrate that the field-induced spin and orbital moments in paramagnetic metals in general are parallel, regardless of the filling of the electronic shell. The early actinides, however, approach the border where the moments go antiparallel. This results in peculiar magnetic states for α-Pu and some uranium compounds, where the spin moments are antiparallel to the applied field and the magnetic response is dominated by the orbital contribution, and consequently these systems display unusual spin densities and magnetic form factors.

We demonstrate that the field induced spin and orbital moments in paramagnetic metals in general are parallel, since the Zeeman energy overcomes the spin-orbit energy that is in favor of an antiparallel arrangement when the electronic shell is less than half-filled. In the early actinides, however, the spin-orbit energy becomes sufficiently strong to approach the border where the moments can couple antiparallel. This results in peculiar magnetic states for α-Pu and some uranium compounds, where the spin moments are antiparallel to the applied field and the magnetic response dominated by the orbital character, and consequently these systems display unusual spin densities and magnetic form factors.

We demonstrate that the conventional picture of crystal structure stabilities of transition metals, namely, as being determined by the degree of filling of the ``canonical'' d bands, breaks down at very high pressures. We show, by means of first-principles calculations, that for extreme compressions the pseudocore p states become broad and start to hybridize with the valence states. This results in a modification of the electronic structure and consequently unexpected crystal structures become stabilized. Simple model calculations using p- and d-canonical bands, which are allowed to hybridize with each other, predict that the bcc structure is the most stable crystallographic phase at high pressure for most of the transition elements, in agreement with our first-principles calculations.

From the experimental behavior of the magnetocrystalline anisotropy energies of the pseudobinary compounds Y(Co1−xFex)5, it has been argued that the magnetocrystalline anisotropy energies for YCo5 and the hypothetical compound YFe5 will have different signs. This anomalous behavior is attributed to the change of the number of 3d electrons and their orbital moments when proceeding from YFe5 to YCo5. The magnetocrystalline anisotropy energies are calculated using the linear muffin-tin orbital (LMTO) method in the atomic sphere approximation (ASA) including spin-orbit interaction and orbital polarization. The force-theorem is used to express the total energy difference (between the two directions of magnetization) as a difference in the sum of the single particle eigenvalues. We find that it is possible to predict the correct easy-axis for YCo5 and YFe5. Secondly it is found that the inclusion of orbital polarization is essential for the cobalt compound but less important for the iron compound. The different contributions from the two inequivalent transition metal sites to the anisotropy energy and orbital magnetization are discussed.

The curie temperatures of the rare earth metals have been calculated ab initio using the local pin density approximation. The exchange splitting of the 5d-states depends upon local 4f-5d exchange integrals, κ415d, which are calculated ab initio in the local spin density approximation. The Curie temperatures may then be obtained without use of adjustable parameters but are calculated to be too high by a factors of two to three if normal itinerant electron mean field theory is used. We have found that it is necessary allow for disordered local 5d moments above Tc and to calculate the local susceptibility. When the local fluctuation contribution to the Landau Parameter, A, is included the Curie temperatures are actually reduced to below those measured.

In the local spin density approximation to the density functional theory the interaction,KRM, between the rare-earth 4f moment and local 4f-5d exchange integrals. Detailed examples are given for the RFe2 (R=Gd-Yb) series. The calculated local exchange integrals are shown to be energy dependent and are then related to the molecular fields.

Electronic structure calculations by means of the LMTO-ASA method have been performed for the hypothetical rare earth-transition metal compound GdFe12 with the ThMn12 structure. The R-4f magnetic moments were obtained from the standard Russel-Saunders scheme but the radial 4f spin density was otherwise part of the self-consistent band calculation. The influence of localized 4f magnetism upon the conduction band magnetism is found to give noticeable changes in the local moments of the iron. The presence of the 4f spin moment is found to induce a redistribution of the conduction electron spin moment between the rare earth and iron sites while the total conduction moment remains practically constant.

The interaction, KRM, between the rare-earth 4f moment and the transition-metal 3d moments in rare-earth transition-metal intermetallics is shown to depend upon the R-5d moment, which is due to 3d–5d hybridization, and local 4f–5d exchange integrals. Both the R-5d moment and KRM may be calculated ab initio from the local spin-density approximation to density functional theory in self-consistent energy-band calculations with the localized 4f-moments fixed at their Russel–Saunders values. Detailed examples are given for the RFe2 (R=Gd−Yb) series. The exchange integrals are similar to those entering into the density functional version of Stoner theory and their energy dependence must be treated carefully. The calculated local exchange integrals are shown to be related to the molecular fields derived from spin Hamiltonians, hence to the spin-wave spectra. Reasonable agreement with values of the molecular fields extracted from inelastic neutron scattering and high field susceptibility measurements is obtained.