Rapid development of information technologies increases the amount of data, which are mainly stored on hard disks made of thin magnetic films. The capacity of hard disks is directly increased by increasing the recording density or, equivalently, by reducing the dimensions of magnetic grains where information bits are stored. The reduction of the volume of particles is limited by the appearance of the superparamagnetic effect, which results in fluctuations of the direction of the magnetization of a grain. A criterion for the stability required to store data on devices has the form KuV/(kBT) ≥ 60, where Ku is the uniaxial anisotropy constant, V is the volume of the grain, kB is the Boltzmann constant, and T is the temperature [1]. The dimensions of magnetic grains can be reduced by using materials with high magnetocrystalline anisotropy such as ferromagnetic FePt thin films in the ordered L10 phase. The magnetocrystalline anisotropy constant in L10-FePt epitaxial films on a MgO (001) substrate reaches 107 erg/cm3 [2, 3]. In addition, these films are easy-axis systems with the equilibrium magnetization perpendicular to the film. On the one hand, a large anisotropy constant allows the reduction of the magnetic grain size, remaining far from the superparamagnetic limit. On the other hand, because of the large anisotropy constant, sometimes unrealistic magnetic fields are required to switch the magnetization. To reduce the switching magnetic field for magnetically hard media, a heat-assisted magnetic recording technology with the heating of a magnetic grain by, e.g., light immediately before the application of the magnetic field to write a bit was proposed [4].

An important parameter of a ferromagnetic material for the heat-assisted magnetic recording technology is the Curie temperature, which affects the time and energy required to remagnetize particles. The introduction of various impurities to the FePt compound makes it possible to control the Curie temperature of the resulting material without a significant change in the magnetocrystalline anisotropy constant. The substitution of nickel [5], manganese [6], and copper [7] for iron is used for such a control. The substitution of rhodium for platinum also reduces the Curie temperature [8].

In this work, we report the study of FePt and FePt0.84Rh0.16 epitaxial thin films with the perpendicular magnetic anisotropy, which indicates that the intense photoexcitation of these materials by femtosecond light pulses leads to the long-lived nonthermal reduction of the coercive field, which is recovered in several milliseconds.

The samples under study were FePt and FePt0.84Rh0.16 continuous epitaxial thin films on MgO (001) substrates; their synthesis method and primary characterization are presented in Section A of the supplementary material. The method and instruments of measurements are described in Section B of the supplementary material. Figure 1 presents static remagnetization curves for the FePt and FePt0.84Rh0.16 films measured by the methods of vibration magnetometry and polarization plane rotation in the magneto-optic Kerr effect in the polar geometry.

Fig. 1.
figure 1

(Color online) Magnetic field dependences of (left axes) the magnetization measured at temperatures of (blue lines) 300 and (red lines) 350 K and (right axes) Kerr rotation angle for the (a) FePt and (b) FePt0.84Rh0.16 samples.

Full size image

The magnetization is saturated in fields of about 0.5 T in both samples. The saturation magnetizations for the FePt and FePt0.84Rh0.16 films at the temperature T = 300 K are 1050 and 700 emu/cm3, which correspond to Kerr rotation angles of 22 and 18 mrad, respectively. The increase in the temperature of the FePt and FePt0.84Rh0.16 samples to 350 K reduces the saturation magnetization by approximately 6 and 10%, respectively. The difference between two materials in the response to heating is due to a noticeably lower Curie temperature TC = 590 K for FePt0.84Rh0.16 versus TC = 740 K for FePt. It is important that the coercive field hardly changes under heating and is about 100 and 89 mT for the FePt and FePt0.84Rh0.16 films, respectively.

The dependence of a change in the Kerr rotation angle after an ultrashort pump pulse with a wavelength of 400 nm and a duration of 50 fs in a magnetic field of B = 0.55 T at various pump energy densities is shown in Fig. 2. The probe pulse had a wavelength of 800 nm and a duration of \( \sim {\kern 1pt} 40\) fs. It is seen that the character of the dependence changes with increasing pump energy density: a fast picosecond relaxation component seen at low pump energy densities is not observed at high values.

Fig. 2.
figure 2

(Color online) Change in the Kerr rotation angle for (a) FePt and (b) FePt0.84Rh0.16 thin films in a magnetic field of 0.55 T perpendicular to the films versus the time delay between the pump and probe pulses at various pump energy densities indicated in panel (a).

Full size image

Further, we report results obtained for the time-resolved magneto-optic Kerr effect. Figures 3a and 3b present the evolution of hysteresis curves for both samples with the time delay between the probe and pump pulses at the pump energy density \(\Phi = \) 6.4 mJ/cm2 and a pulse repetition rate of 1 kHz. The maximum suppression of the magnetization observed in the time delay range Δt = 2–50 ps reaches 72 and 88% for the FePt for FePt0.84Rh0.16 films, respectively. At long time delays, the saturation magnetization is recovered in 1.7 ns to almost the initial value. In the presence of long-term dynamics, a small negative time delay \(\Delta t = - 10\) ps can be considered as a large positive time delay equal to the laser pulse repetition period Tl with respect to the preceding pump pulse (Δt = Tl). As seen in Figs. 3a and 3b, the amplitude of the hysteresis curve at this time delay is completely recovered.

Fig. 3.
figure 3

(Color online) Remagnetization curves in the form of the magnetic field dependences of the Kerr rotation angle for (a, c) FePt and (b, d) FePt0.84Rh0.16 epitaxial thin films at the temperature T = 300 K, a pump energy density of 6.4 mJ/cm2, and time delays between the probe and pump pulses Δt = (blue lines) 10, (light blue lines) 200, and (orange lines) 1700 ps. The dashed lines are the equilibrium curves. Time-resolved data are presented for repetition rates of pump pulses of (a, b) 1 and (c, d) 0.1 kHz. The dotted lines present data for \(\Delta t = - 10\) ps, which corresponds to the delay time from the preceding pulse of (a, b) 1 and (c, d) 10 ms.

Full size image

The most outstanding feature of the hysteresis loops for both samples under pulsed photoexcitation is the time-stable suppression of the coercive field by about 70 and 83% for FePt and FePt0.84Rh0.16, respectively. Coercivity remains low even in 1 ms after the pump pulse. To experimentally estimate how long coercivity is reduced, similar measurements were carried out at a pulse repetition frequency of 100 Hz. Their results are shown in Figs. 3с and 3d. In this case, coercivity is also reduced stably in the entire time delay range but this suppression is less pronounced. The coercive field for the FePt film is reduced insignificantly, maximum by \( \sim {\kern 1pt} 18{\kern 1pt} \% \), whereas the corresponding reduction for the FePt0.84Rh0.16 film is \( \sim {\kern 1pt} 60{\kern 1pt} \% \). An additional difference of these measurements from experiments at a laser frequency of 1 kHz is a weaker photoinduced decrease in the magnetization at all time delays and the same pump energy density. The maximum change in the saturation magnetization in the FePt and FePt0.84Rh0.16 samples is 45 and 78%, respectively.

The dependences of the coercive field of the samples on the laser pulse repetition period Tl and on the pump energy density are presented in more detail in Fig. 4. The hysteresis loop was measured in these experiments in the rotation angle of the polarization plane at the time delay of the probe pulse \(\Delta t = - 10\) ps (equivalent to Δt = Tl, as mentioned above). Correspondingly, Fig. 4a illustrates the long-term (up to 50 ms) dynamics of the recovery of the coercive field (at a pump energy density of 5.5 mJ/cm2). The solid lines in Fig. 4а are the fits of the data by the function \(y = A - B\exp ( - {{T}_{l}}{\text{/}}\tau )\) with the coercivity recovery time constants τ = (\(1.86 \pm 0.15\)) ms and τ = (\(2.8 \pm \) 0.3) ms for FePt and FePt0.84Rh0.16 thin films, respectively.

Fig. 4.
figure 4

(Color online) Coercive field for the (squares) FePt and (circles) FePt0.84Rh0.16 measured at a negative time delay of \( - 10\) ps (see the main text) (а) versus the repetition period at the pump energy density \(\Phi = 5.5\) mJ/cm2 and (b) versus the pump energy density at the repetition frequency of pump pulses f = 1 kHz. The lines in panel (а) are approximations by the function \(y = A - B\exp ( - {{T}_{l}}{\text{/}}\tau )\).

Full size image

Figure 4b shows the dependence of the coercive field of the studied thin films on the pump energy density at a frequency of 1 kHz. The coercive field for both samples decreases with increasing pump energy density though the characters of these dependences are somewhat different for the FePt and FePt0.84Rh0.16 films. The coercive field in the FePt film decreases gradually in the entire range with an increase in the pump energy density and reaches a minimum value of 24 mT at a pump energy density of 6.4 mJ/cm2. The decrease in the coercive field in the FePt0.84Rh0.16 film with an increase in the pump energy density is weak in the ranges Φ = 0–3.2 mJ/cm2 and Φ > 5 mJ/cm2 and is strong in the range Φ = 3.2–4.1 mJ/cm2.

Thus, a strong dependence of the coercive field on the pump pulse repetition period is manifested in experiments on the ultrafast photoinduced dynamics of the magnetization of continuous epitaxial ferromagnetic thin films with the structure of the tetragonal L10 phase and perpendicular magnetic anisotropy. Interestingly, the state with the reduced coercivity holds for a fairly long time of milliseconds. In this case, the magnetization is recovered after the pump pulse to the initial value already in 2 ns and hardly changes after that. At the same time, the static heating of films from 300 to 350 K (Fig. 1) primarily reduces the magnetization with the conservation of the coercive field (this effect is more pronounced for the FePt0.84Rh0.16 film). This behavior indicates the nonthermal nature of the photoinduced reduction of the coercive field. However, at pump pulse repetition rates of 1 kHz and less, the heat accumulation effect is usually negligibly small, and the initial equilibrium temperature is recovered in nanoseconds.

A long recovery time of the coercive field is remarkable. It could be assumed that the observed reduction is due to macroscopic phenomena, e.g., a photoinduced structural phase transition. Indeed, an increase in the symmetry of the structure to the cubic one could significantly reduce the coercive field. However, the gradual variation of the coercive field without jumps with a decrease in the pulse repetition period (Fig. 4а) shows that our observations are most likely not due to the photoinduced phase transition.

Furthermore, if the heating of the film resulted in the structural phase transition from the tetragonal \(L{{1}_{0}}\) phase to the high-temperature cubic А1 phase (see, e.g., [9]), the accumulation of structural disordering of the low-temperature phase would be expected at a laser pulse repetition rate of 1 kHz because the energies of the ordered L10 phase and the disordered cubic phase of the same composition are very close to each other. This disordering would be manifested in a decrease in the equilibrium coercive field whose value in the disordered phase is much smaller than that in the tetragonal L10 phase (see, e.g., [10]). Meanwhile, the equilibrium coercive field in our experiment is independent of its duration.

Finally, a nonthermal mechanism of the reduction of the coercive field is additionally confirmed by the dynamics of the magnetization, which is recovered in about 1.7 ns after the pump pulse and then remains unchanged. At the same time, the coercive field remains reduced to \( \sim {\kern 1pt} 2\) ms, i.e., to times six orders of magnitude longer than the magnetization recovery time.

Magnetization reversal of continuous thin films with perpendicular magnetic anisotropy in the magnetic field applied along the easy axis (Figs. 1 and 3) include two necessary stages: first, the nucleation of domains with the magnetization opposite to the initial magnetization and, second, their growth due to the motion of domain walls. The former stage is manifested in a bend in the hysteresis loop with a decrease in the absolute value of the magnetization compared to the saturated state. “Friction” caused by the motion of domain walls at the growth of domains is related to their pinning to defects in the film. The pinning energy scale determines the coercive field at a given temperature. The experimental results presented in Fig. 3 certainly indicate that both stages of remagnetization in both FePt and FePt0.84Rh0.16 films occur under the pulsed photoexcitation at a much lower magnetic field, which generally leads to significant magnetic softening. Moreover, since heating is absent during the overwhelming fraction of the time interval between pump pulses, the observed magnetic softening has a nonthermal nature. What can be a reason different from heating?

Both nucleation and motion of domain walls have an activation nature. Therefore, the softening of the magnetic characteristics in the absence of heating requires either a decrease in the corresponding activation energies or the “nonthermal pumping” of the energy to the magnetic subsystem. The authors of [11] reported that the coercive field in an epitaxial film of the Ga1 – xMnxAs0.96P0.04 magnetic semiconductor with perpendicular anisotropy was reduced by a factor of 2 under the action of the ultrasonic pulses exciting surface acoustic waves. The ultrasonic frequency, pulse duration, and repetition period were 549 MHz, 600 ns, and 20 ms, respectively. The authors of [11] attributed the observed effect to the periodic decrease in the domain nucleation energy, which depends on the strain, under the action of surface acoustic waves. Related phenomena associated with the laser-induced modification of magnetic anisotropy were discussed in [12]. The current status of ultrafast magneto-acoustics was reviewed in [13].

The phenomenon described in this work can have a similar nature. First, it is known that the absorption of ultrashort light pulses leads to the generation of a hypersonic wave with a wide (up to hundreds gigahertz) frequency spectrum [14, 15]. Second, the FePt compound is characterized by the record magnetoelastic coupling: magnetostriction in this compound is even higher than that in nickel [16–18]. Thus, the acoustic wave induced by a femtosecond light pulse can increase the energy of the magnetic subsystem. Third, resonant acoustic modes in samples of a regular shape such as a plane–parallel single crystal MgO substrate can have a high Q factor [19] and, correspondingly, can serve as a low-dissipation energy reservoir with an efficient energy transfer channel to the magnetic film. The epitaxial character of the studied films ensures in turn strain transfer from the substrate to the film, which can increase the energy of the magnetic subsystem and facilitate both the nucleation of domains and the nonthermal depinning of domain walls due to the strong magnetoelastic coupling. A high Q factor of acoustic resonances is responsible for the long-term reduction of the coercive field. In our opinion, the difference between the recovery times of the coercive field for the two compositions of the films may be due to the difference between the magnetoelastic interaction constants. The spin–orbit coupling in rhodium is weaker because it is a noticeably lighter element, and the substitution of rhodium for platinum should thereby reduce the magnetoelastic constant.

A similar experimental manifestation of the decrease in the switching field of the magnetic Josephson junction under microwave pumping was observed in [20]. The result was interpreted as the microwave excitation of fluctuations of the local magnetic moments of PdFe clusters, which reduces the coercive field and simplifies the remagnetization process.

To summarize, our studies of FePt and FePt0.84Rh0.16 epitaxial thin films on MgO single-crystal substrates have revealed the nonthermal effect of photoinduced magnetic softening, which leads to the long-term reduction of the coercive field. It has been proposed to explain the effect by the excitation of femtosecond laser pulses of acoustic resonances in the substrate–film system. We believe that this effect can serve as a foundation for a technology alternative to heat-assisted magnetic recording used in magnetic hard disks based on FePt granular media.