Testing spin-flip scattering as a possible mechanism of ultrafast demagnetization in ordered magnetic alloys

We use element-resolved IR-pump/extreme ultraviolet-probe experiments to disentangle the ultrafast interplay of the magnetic sublattices of an ordered crystalline magnetic alloy. As a paradigmatic example, we investigate the case of the FeRh alloy, which shows a delayed response for the different components. Furthermore, a detailed time-resolved magneto-optic study shows that the data can be analyzed by only assuming Elliot-Yafet-like scattering, as the underlying mechanism for ultrafast demagnetization, resulting in an unexpected nonmonotonic dependence of the spin-flip rate, as a function of quenching.

Figure 1

Sketch of the CITIUS HHG light source. The 3 mJ/pulse Ti:sapphire laser output is split in two branches: 2 mJ/pulse is used for HHG and 1 mJ/pulse for sample excitation (IR pump). The proper harmonic is selected by means of a time preserving monochromator based on conical diffraction gratings (CDGs) and refocalized onto the sample. The IR pump is recombined with the HHG light in the experimental chamber IRMA on the sample in an almost collinear geometry. (a) The higher harmonics spectrum from Ne after the monochromator captured with a photodiode and (b) the normalized magnetic asymmetry ratio (A.R.), namely, the difference of both field directions divided by their sum, for the T-MOKE-like configuration in the IRMA setup.

Figure 2

Element-resolved demagnetization dynamics of FeRh (Fe: blue open squares; Rh: red open circles) obtained by TR-XUV-TMOKE. The reduced magnetization ($M/M_0$) after excitation with (a) 11 mJ/${\mathrm{cm}}^2$ and (b) 16 mJ/${\mathrm{cm}}^2$. The inset in (b) shows a zoom to the initial drop of the signal. Fe demagnetization is delayed with respect to Rh by approximately 60 fs.

Figure 3

TR-MOKE demagnetization traces of FeRh for various fluences (0.4–10 mJ/${\mathrm{cm}}^2$). The solid line (red) is a fit using the M3TM model, where only the spin-flip rate and the excitation strength is allowed to vary. The demagnetization time $\tau$ (dotted line) referenced to the quenching $q$ (defined as the loss of magnetization at 1 ps, dashed-dotted line) is visualized for the trace with the highest fluence (10 mJ/${\mathrm{cm}}^2$).

Figure 4

Fit parameters from the M3TM for element-resolved measurements, Rh (red open circles), Fe (blue open squares), and for TR-MOKE and the transient optical response (black open triangles; error bars are within the symbol size). (a) The demagnetization time $\tau$ and (b) the spin-flip rate $a_{\text{sf}}$, which is a key parameter for describing the demagnetization time and the quenching vs fluence. (c) The difference in $t_0$ for the element-resolved demagnetization times shows that Fe starts delayed by about 60 fs with respect to Rh for various fluences. Error bars for the lower fluence appear bigger due to a lower signal and a nearly constant noise level in all measurements. (d) The electron-phonon coupling $g_{ep}$ is not constant as a function of the maximum electronic temperature or the resulting calculated quenching.