AVS 45th International Symposium
    Magnetic Interfaces and Nanostructures Technical Group Thursday Sessions
       Session MI-ThM

Paper MI-ThM4
On the Nature of Resonant Photoemission in Gd

Thursday, November 5, 1998, 9:20 am, Room 324/325

Session: Magnetic Spectroscopies
Presenter: J.G. Tobin, Lawrence Livermore National Laboratory
Authors: J.G. Tobin, Lawrence Livermore National Laboratory
K.W. Goodman, Lawrence Livermore National Laboratory
S.R. Mishra, Virginia Commonwealth University
W.J. Gammon, Virginia Commonwealth University
T.R. Cummins, University of Missouri, Rolla
G.D. Waddill, University of Missouri, Rolla
G. van der Laan, Daresbury Laboratory, England
Correspondent: Click to Email
The phenomenon of "resonant photoemission" occurs when, in addition to a direct photoemission channel, a second indirect channel opens up as the absorption threshold of a core level is crossed. A massive increase in emission cross section can occur, but the nature of the process remains clouded. Is it truly "resonant photoemission" or merely the incoherent addition of a second emission channel? Using novel magnetic linear dichroism in photoelectron spectroscopy experiments and computational simulations, we can now clearly demonstrate that temporal matching of the processes as well as energy matching is a requirement for true "resonant photoemission." The photoemission of 4f and 5p electrons from rare-earth metals and their compounds is strongly enhanced when the photon has just enough energy to excite a 4d electron to an unoccupied 4f level, leading to a process called "resonant photoemission". In a generic picture, the indirect channel of the resonant photoemission is interpreted as due to a process where a 4d electron in the initial state is first excited to the unoccupied 4f level, forming a tightly coupled, bound intermediate state, 4d core hole plus 4f electrons. Then a decay via autoionization occurs into the final state, thus producing a final state indentical to that obtained by a direct photoemission process for the ejected electron. The transition rate is greatly enhanced if the excited state decay is by a Coster-Kronig or a super-Coster-Kronig process. The key question is whether these processes are coherent or incoherent: should the overall intensity be treated as a squaring of the sum of the amplitudes (coherent) or summing of the squares of the amplitudes (incoherent)? A true "resonant photoemission" process should be coherent, involving interference terms between the direct photoemission and indirect photoemission channels. Possibly, incoherence would give rise to the loss of photoemission characteristics in the process, with a domination of auger-like properties. To this problem we have applied the new photoelectron spectroscopy technique of magnetic linear dichroism in angular distributions (MLDAD). This technique is related to but distinct from the techniques of magnetic xray circular dichroism (MXCD) in photoelectron spectroscopy and xray absorption. The key is that while large dichroic effects in ferromagnets can be observed with MXCD-photoemission and MXCD-absorption, the large MLDAD effects in ferromagnets is solely a photoemission, not an absorption-driven, process. This is because the chirality which gives rise to magnetic sensitivity is due to the vectorial configuration in MLDAD as opposed to the intrinsic chirality of circularly polarized xrays in the MXCD techniques. In absorption, where there is an essential averaging over all emission angles, the vectorial chirality is lost. Thus, MLDAD is the perfect measurement to distinguish between photoemission and absorption processes. Angle-resolved photoemission in a magnetic system should show an MLDAD effect: xray absorption and thus auger emission will show no MLDAD effect. It is this test which we have applied to the "resonant photoemission" of the Gd5p and Gd4f emissions.