Experimental energy band dispersions and magnetic exchange splittings for Fe, Co and Ni

Abstract

A long-standing goal has been the determination of accurate exchange-split band dispersions for the itinerant ferromagnets Fe, Co and Ni. As often noted, magnetic exchange splittings and band dispersions E( k) are fundamental for understanding various physical properties involving d-band electronic structure, collective itinerant-electron ferromagnetism, transition metal surfaces, etc. Using polarization-dependent angle-resolved photoemission and synchroton radiation, we have recently determined the temperature-dependent exchange splitting and band dispersions for Ni [1, 2], Fe [3] and Co [3]. We have shown that direct transitions are of primary importance for normal photoemission from Ni(111), Fe(111) and Co(0001) and determine accurate exchange-split band dispersions E( k) for fcc Ni along the Γ symmetry line, for bcc Fe along the ΓPH symmetry line and for hcp Co along the ΓAΓ line. These results together with experimental E( k) dispersions for Cu [4] permit systematic comparison with state-of-the-art ab-initio band calculations [5]. This comparison shows that the ratio of the theoretical-to-experimental occupied d-band width is about 1.1, 1.2, 1.45 and 1.1 for Fe, Co, Ni and Cu, respectively, while the ratio of the theoretical-to- experimental exchange splitting δ E ex is about 1.0, 1.2 and 2.2 for Fe, Co and Ni, respectively. General conclusions are as follows: (1) Fe, Co and Ni can all be described by a Stoner-Wohlfarth-Slater (SWS) spin-split band model, and (2) state-of-the-art ab-initio one-electron band calculations quantitatively described Fe and Cu quite accurately, Co less well and Ni rather poorly. This behavior is consistent with electron-electron correlation effects for unfilled d-shells which increase in importance on going from Fe to Co to Ni.