We compare femtosecond pump-probe experiments in Ni and micromagnetic modelling based on the Landau-Lifshitz-Bloch equation coupled to a two-temperature model, revealing a predominant thermal ultrafast demagnetization mechanism. We show that both spi
n (femtosecond demagnetization) and electron-phonon (magnetization recovery) rates in Ni increase as a function of the laser pump fluence. The slowing down for high fluences arises from the increased longitudinal relaxation time.
The use of femtosecond laser pulses generated by a Ti:Sapphire laser system allows us to gain an insight into the magnetization dynamics on time scales from sub-picosecond up to 1 ns directly in the time domain. This experimental technique is used to
excite a polycrystalline nickel (Ni) film optically and probe the dynamics afterwards. Different spin wave modes (the Kittel mode, perpendicular standing spin-wave modes (PSSW) and dipolar spin-wave modes (Damon-Eshbach modes)) are identified as the Ni thickness is increased. The Kittel mode allows determination of the Gilbert damping parameter alpha extracted from the magnetization relaxation time tau_alpha. The non-local damping by spin currents emitted into a non-magnetic metallic layer of vanadium (V), palladium (Pd) and the rare earth dysprosium (Dy) are studied for wedge-shaped Ni films 1 nm-30 nm. The damping parameter increases from alpha=0.045 intrinsic for nickel to alpha>0.10 for the heavy materials, such as Pd and Dy, for the thinnest Ni films below 10 nm thickness. Also, for the thinnest reference Ni film thickness, an increased magnetic damping below 4 nm is observed. The origin of this increase is discussed within the framework of line broadening by locally different precessional frequencies within the laser spot region.
By means of time-resolved Kerr spectroscopy experiments we relate the energy dissipation processes on the femtosecond (electron-spin relaxation time $tau_{el-sp}$) and nanosecond timescale (Gilbert relaxation $tau_{alpha}$) and compare the results to
the first microscopic model, which was proposed by Koopmans. For both energy dissipation processes, Elliot-Yafet scattering is proposed as the dominant contributor. We controllably manipulate the energy dissipation processes by transition metal doping (Pd) and rare earth doping (Dy) of a Permalloy film and find that while a change of $tau_{alpha}$ of more than a factor two is observed, tau_{el-sp}$ remains constant, contrary to the predictions of the model. We explain the discrepancies by relaxation channels not considered in the original microscopic model and identify thereby the applicability of the model and possible necessary extensions to the model.
