Ultrafast hot-electron induced quenching of Tb $4f$ magnetic order

We have investigated ultrafast quenching of the Tb $4f$ magnetic order in $\mathrm{C}{\mathrm{o}}_{74}\mathrm{T}{\mathrm{b}}_{26}$ alloys, induced by femtosecond hot-electron pulses. The hot-electron pulses were produced in specific nonmagnetic capping layers by infrared femtosecond laser pulses. Our experimental results show that subpicosecond dynamics of Tb $4f$ magnetic moments can be induced by nonthermal and thermal hot-electron. We further demonstrate that the demagnetization efficiencies of nonthermal and thermal hot-electron are similar. However, the characteristic demagnetization times show values of 0.35 ps for nonthermal hot-electron excitations and 1.2 ps for thermal hot-electron excitations. We explain this temporal elongation by the propagation time of thermal hot-electron through the 15-nm-thick CoTb film.

Figure 1

(a) Sample 1 configuration used as the reference sample and pumped with the IR laser. (b) Sample configurations for sample 2 used to excite the CoTb film with ballistic hot-electron. (c) Sample configurations for sample 3 used to excite the CoTb film with diffusive electrons.

Figure 2

(a) XMCD spectra at the $\mathrm{Tb}{M}_5$ edge and (b) hysteresis loops for the samples 1, 2, and 3. The measurements were carried out at $T=300\mathrm{K}$ for a normal incident of the x-ray beam in respect with the sample's surface. XMCD spectra are defined as the difference of XAS spectra obtained for two magnetic field helicities ($H=0.7\mathrm{T}$). For all samples, the spectra are normalized, such as the amplitude of the isotropic XAS spectrum at the $M_5$ resonance ($E_{\mathrm{hv}}=1237\mathrm{eV}$) is equal to 1 (the baseline at $E_{\mathrm{hv}}=1220\mathrm{eV}$ is equal to 0). Hysteresis loops were acquired by monitoring the x-ray transmission at the Tb $M_5$ resonance ($E_{\mathrm{hv}}=1237\mathrm{eV}$) as a function of the magnetic field.

Figure 3

Transient normalized XMCD at the $\mathrm{Tb}{M}_5$ edge for samples 1, 2, and 3. For samples 2 and 3, the solid lines are fits in the framework of the three-temperature model. For sample 1, which is used to set the temporal overlap in this experiment, we superpose the data (open circles) and the fit of our previous $\mathrm{Tb}{M}_5$ dynamics taken in the same experimental configuration [33]. The different sample configurations (samples 1, 2, and 3) show the respective characteristic demagnetization times of $\tau =0.28\pm 0.03\mathrm{ps}$, $0.35\pm 0.1\mathrm{ps}$, and $1.2\pm 0.4\mathrm{ps}$. The data in Fig. 3 (top panel) are reproduced with the authorization of the American Physical Society.