Research highlights
Structural Dynamics during Ultrafast Demagnetization (June 2017)
Most experiments focused so far on characterizing magnetization and charge carrier dynamics, while the first direct measurements of structural dynamics during ultrafast demagnetization were reported only very recently. We here present our investigation of the infrared laser-pulse-induced ultrafast demagnetization process in a thin Ni film, which characterizes simultaneously magnetization and structural dynamics. To bring new information we have employed resonant magnetic x-ray reflectivity to follow simultaneously magnetization and structural dynamics at the BESSY femtoslicing source. We find that significant changes in nonmagnetic x-ray reflectivity accompany the subpicosecond demagnetization, which can be modeled as a variation of film thickness. We are further investigating what is at the origin of an ultrafast contraction and if this phenomenon depend on the magnetic layer and/or cap layer.
Figure 1: TR-XRMR principle and non-magnetic reflected intensity and normalized magnetization in function of time delay for a reflected angle of 10.9°. In red, Ni thickness in function of time delay for the upper pannel, and exponential fitting for the lower pannel.
Our simulations further show that the higher photon flux and energy resolution provided by x-ray free electron lasers will yield decisive data to differentiate between different mechanisms proposed to govern ultrafast demagnetization dynamics. Indeed, The most broadly accepted model is based on Elliot-Yafet like electron-phonon scattering, but over the past years experimental evidence has been given for the occurrence of angular momentum transport by the excited, spin polarized valence electrons. Battiato et al. showed that such a transport occurring in the regime of superdiffusion is sufficient to explain the ultrafast demagnetization. Further evidence has been found, but these results are discussed controversial. More recent studies to explain the ultrafast magnetization dynamics suggest the co-existence of both local and non-local processes. We note, however, that so far no direct experimental proof has been obtained for the existence of a superdiffusive spin transport in a single magnetic thin film as predicted by theory.
However, the two models predict distinctly different developments of the magnetization in the direction perpendicular to the film surface. Elliott-Yafet scattering will reduce the magnetization locally and the shape of the magnetization depth profile will be given by the depth dependence of the excitation process, i.e., the depth profile of the absorbed IR pulse intensity. Spin-polarized electron transport as a non-local process, on the other hand, will give rise to a very specific evolution of this excitation profile as predicted by Battiato et al. From our experimental results describe above and thanks to simulations, we can show that based on the photon beam parameters of an XFEL there are distinct differences clearly indicating that the predictions of the superdiffusive spin transport model can be tested in a quantitative manner. We therefore propose to investigate the magnetization amplitude in time and space during ultrafast demagnetization with a pump-probe XRMR experiment at XFELs.
Figure 2 : Experimental (green squares) and calculated magnetization asymmetry for the 3 demagnetization profiles homogeneous (black), Elliott-Yaget (blue) and Superdiffusive (red) and for slicing condition in a) and XFEL conditions in b) and c).