Spin Polarized Electron Scattering by Ferromagnetic Amorphous Surfaces'

It is shown that kinematic atomic scattering is the predominant mechanism for the elastic backward . scattering of electrons from the ferromagnetic glass Fe40Ni40B20 as opposed to the resonance model recently suggested by R. K. Nesbet. The spin dependence of this scattering was determined theoretically (in the energy range below 300 eV) using various spin polarized potentials for Fe and Ni. Below 30 eV, current exchange correlation models fail to explain the experimental data even when allowing for a spin dependence in the attenuation of electrons. (Submitted to Physical Review Letters) * Work supported most by Department of Energy under Contract DE-AC03-76SFOO515 and in part by DE-AC03-76SF00098. ** Present address: Department of Chemistry, University of California Berkeley, CA 94720 Scattering of slow spin polarized electrons from magnetic surfaces emerged recently as one of the most promising techniques to study surface magnetism l-4 . The surface magnetism can be detected by measuring the scattering asymmetry factor S defined as S = 1+ I1++ I-' in which I + and Iare the scattered intensities of the incident electrons having spin directions polarized parallel or antiparallel to the majority spin of the sample. A comparison of the dependence of S on the incident electron energy E between magnetic single crystal surfaces and amorphous surfaces appears to be very interesting. Wang3 has shown that S-E curves for single crystal surfaces show rich structures due to a superposition of low energy electron diffraction with the structures due to surface magnetism whereas Schilling and Webb5 have shown that in the case of electron scattering from liquid Fig5 the intensity due to atomic scattering can be separated from the effects of crystal diffraction and multiple scattering within reasonable approximations. Pierce et al6 recently measured the S-E curves from an amorphous ferromagnet Fe 40Ni40B20' These curves appear much simpler than those from a single crystal Fe surface and change sign only near 10 eV and 50 eV. Between these two energies S is negative, i.e., electrons polarized antiparallel to the sample spin are preferentially scattered. S is otherwise positive throughout the measured energy region (~300 eV). Recently Nesbet' proposed an electron resonance model to explain the change of sign near 50 eV. Namely, in Fe atom the electronic configuration is 3p64s23d6 and so long as two localized d-shell hole states are vacant, an electron incident'on Fe with sufficient energy E* for a 3p -t 3d excitation could be temporarily captured into a resonance state of Fewith configuration 3p 5 2 8 4s 3d . Since the magnetic moment of Fe atom is 2 uB (Bohr magneton), the resonance state could be a doublet (if the incident electron spin is'-') or quartet (if the incident electron spin is '+')depending on the polarization of the incident electron.