Femtosecond laser-heating effect on the magnetization dynamics in perpendicularly magnetized Ta/CoFeB/MgO film

We have studied the effect of ultrafast laser-heating on the magnetization dynamics of perpendicularly magnetized CoFeB film by means of the time-resolved magneto-optical Kerr rotation effect. The effective perpendicular magnetic anisotropy field H K is significantly decreased with enhancing the pump laser-fluence in a moderate range of 5–12 mJ cm−2. The Gilbert damping, however, is found to be independent of the pump fluence. These findings provide a new method of separately manipulating the Gilbert damping and perpendicular magnetic anisotropy.


Introduction
The magnetic films with perpendicular magnetic anisotropy (PMA) are the promising candidates in the application of the next-generation, high-density magnetic information storage technology due to their high thermal stability. One of the prevailing magnetic random access memories (MRAM) is switched by spin transfer torque (STT) due to its good down-scalability [1,2]. With STT, the critical current density for magnetization switching in the case of perpendicular magnetized magnetic tunneling junctions (MTJs) is directly proportional to the Gilbert damping constant and PMA. This proportionality shows that a balance between the thermal stability and writability of magnetization is required. Therefore, materials with large PMA and weak Gilbert damping are preferred. The CoFeB/MgO/CoFeB MTJ with PMA is extensively investigated due to its high tunneling magnetoresistance ratio [3]. However, from a physical point of view, magnetic anisotropy and Gilbert damping both arise from the same origin, namely the spin-orbit coupling, as evidenced by previous results [4,5]. Usually, materials with large PMA exhibit large damping values such as Co/Pt multilayers [6] and CoCrPt alloys [7].
At the same time, the PMA and the Gilbert damping can also be influenced by various other factors, such as capping layer, [8][9][10] annealing temperature, [11,12] element compositions, [5,13,14] and film thickness [15,16]. For instance, Liu et al [10] observed a 35% increase in PMA by replacing the Ta buffer layer, with Devolder et al [17] reported that, similar to the annealing effect, the effective anisotropy field increased in proportion to the irradiation fluence of light ions, while the Gilbert damping constant was almost invariant. These studies suggest that there is a possibility of tuning the anisotropy and damping constant in a separate way. In the practical application of STT-MRAM, the magnetization switching generally takes place within a timescale of nanoseconds (ns) and thermal effect arising from the writing current is inevitably introduced into a given device. It remains an open question that how the transient thermal effect influences the anisotropy and Gilbert Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence.
Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
damping. It has been previously reported that the damping constants in ferromagnetic thin films with in-plane magnetic anisotropy can be transiently enhanced by increasing the fluence of femtosecond laser-pulses [18,19]. The results showed that the transient thermal effect induced by the ultrafast laser-pulses within the ns timescale plays an important role in transiently manipulating damping parameter. However, studies of transiently tailoring the PMA and Gilbert damping by this kind of thermal effect are still lacking. In this letter, we report the effect of transient laser heating on the magnetic anisotropy and Gilbert damping in perpendicularly magnetized Ta/CoFeB/MgO film by using the time-resolved magneto-optical Kerr effect (TRMOKE) techniques. The effective perpendicular anisotropy field H K has been found to be strongly dependent on the fluence of the ultrafast laser-pulses. The Gilbert damping, however, remains stable within a pump fluence range of 5-12 mJ cm −2 where the saturation magnetization remains constant.

Experimental methods
The sample was prepared by using the magnetron sputtering method with a base pressure of 10 −5 Pa. After the sample was deposited, a post-annealing process at 300°C was performed. Further growth information can be found in our previous paper [20]. The stacking structure is Si substrate/Ta(5 nm)/Co 40 Fe 40 B 20 (1 nm)/MgO(3 nm)/Ta(5 nm). The numbers in brackets represent film thickness. The magnetization dynamics was measured using the TRMOKE at room temperature. In our measurements, the laser pulses were generated by a Ti: sapphire regenerative amplifier with a central wavelength of 800 nm, a repetition rate of 1 kHz, and a pulse duration of ∼50fs. The output of the amplifier was divided into two beams. One was used as the pump beam and the other was frequency-doubled (3.1eV) as the probe beam. The spot diameters of the pump and probe beams focused onto the sample are ∼500 and 200 μm, respectively. Figure 1(a) shows the experimental geometry employed in our measurements. The pump beam is incident normally and the incident angle of probe beam is ∼5°with respect to the normal direction of the sample. The applied magnetic field H is along the direction of H q =47.5°. Figure 1 (b) shows the hysteresis loops with the magnetic field parallel and perpendicular to the film plane. These magnetization curves were measured by a quantum design superconducting quantum interference device. The saturation magnetization M s is around 1500 emu/cc, which coincides with other results [21,22]. The experimental H K is approximately 3 kOe obtained from the closed area between the in-plane and out-of-plane hysteresis loops.  D which were measured at the pump fluence of 7 mJ cm −2 under different magnetic fields. The instantaneous decrease of Kerr signal at t D =0 ps is characterized as the ultrafast demagnetization [23]. This ultrafast demagnetization behavior modifies the effective magnetic anisotropy field within a few ps and thus triggers a magnetization precession which last for hundreds of ps. The oscillations shown in figure 2(a) correspond to the magnetization precession that strongly depends on the external field H. In addition, the measured data of Kerr rotations also include a non-oscillating component, which can be described by a single-exponential function. Thus, the experimental Kerr dynamics can be modeled as the following equation [24]:

Results and discussions
where a τ, f, f are the initial amplitude, spin relaxation time, precessional frequency and phase, respectively. b is the background amplitude and v is the recovery rate of the magnetization. In figure 2(a), the red solid lines are the fitting results. The extracted parameter of precession frequency f is plotted in figure 2(b) as a function of H. We reproduce f using the following equations as used in previous results [25,26]: where K U is the uniaxial magnetic anisotropy constant. The equilibrium magnetization angle θ is calculated by the equation: The value of H K =2kOe at 7 mJ cm −2 is obtained by fitting the extracted precession frequencies f in  In addition, as shown in previous results [18,19], the variation of M S is negligible for analyzing the pump-fluence dependence of the effective magnetic anisotropic field. Therefore, the PMA K U is proportional to H . K The pump fluence dependent results of K U and 0 a demonstrate that the perpendicular interfacial anisotropy can be varied independently with damping constant. Such a separate tunability cannot be simultaneously achieved in most cases [27,28] because ferromagnetic materials with large anisotropy constant usually have large damping factor due to spin-orbital coupling (SOC). C / Here T is the electronic temperature and T C means Curie temperature. However, in the case of 1 nm thick CoFeB film, besides the similar electron temperature rise under different pump fluences, the interfacial effect has to be taken into consideration. The origin of PMA for the 1nm CoFeB film is attributed to the hybridization of Fe and oxygen's orbitals at the CoFeB/MgO interface due to the interfacial SOC [29,30]. As a consequence, the decrease of PMA as increasing the pump fluence indicates the loss of interfacial SOC. We attribute this reduction of interfacial SOC to the transient thermal effect induced by the pulsed laser excitations. Similar quenching of PMA caused by high temperature was also reported in previous works [31]. Based on the discussions above, in the PMA CoFeB film, both the interfacial SOC and the T T C / ratio are obviously modulated by the thermal effect caused by the pulsed laser. These two factors should be responsible for the different behaviors of H K and 0 a of the 1nm thin film case.
Before showing how the two factors lead to our experimental results, it is essential to approximately quantify the transient thermal effect. Illuminated by the laser pulses of different fluence, the corresponding equilibrium temperatures of the sample was estimated to be ∼815-1310 K. The temperatures were obtained according to the ratios of the reduced magnetic order 0 q q D / at t D =5ps (see figure 4) as well as the published M-T curves [32].
Here, we used the values of The damping term ratio a originates from the bulk properties of the film. As a consequence, to assure the fluence-independent nature of , 0 a another damping term is required to compensate the increase of . ratio a Considering the proportional correlation between the SOC and damping parameter, it is reasonable to define a damping term inter a that arises from the interfacial SOC. As also recently demonstrated by Okada et al [35], the interfacial PMA is found to be correlated with the Gilbert damping manifested by monitoring the ferromagnetic resonance linewidths with varying temperatures and sample thickness. According to the

Conclusion
In conclusion, the femtosecond laser-heating effect on the magnetization dynamics in the perpendicularly magnetized CoFeB film is investigated by the TRMOKE techniques. The equilibrium temperature of the CoFeB film with PMA is quantitatively scaled from 815 to 1310 K in the pump fluence range of 5-12 mJ cm −2 . Within such a high temperature regime, the effective perpendicular anisotropy field decreases with increasing the pump fluence while the Gilbert damping constant remains stable. The possible reasons for fluence-independent behavior of 0 a are discussed in terms of the interfacial SOC and electronic temperature. Our finding provides an approach for controlling separately the perpendicular anisotropy field and the damping constant, which is critical for achieving the best performance of STT-MRAM devices. H q =  The red solid lines are the fitting lines and the details of fitting procedure are given in SM. The fitting equation is from previously published [33,34].