Coherent control of photomagnetic back-switching by double-pump laser pulses

The control of nonthermal all-optical magnetization switching, under a regime of a train of identical laser pulses, opens up opportunities for ultrafast magnetic recording. Here, we investigate the photomagnetic back-switching capabilities of the write and erase magnetic domain pattern using double-pump pulse excitations in an iron-garnet film with pure cubic magnetocrystalline symmetry. It is essential to note that forward and backward magnetization switching is achievable in two distinctive scenarios: using identical linearly polarized laser pulses and with pulses having orthogonal polarization planes. By observing the switch of magnetization at domains independent of the initial state, one can nonthermally toggle the magnetization, equivalent to the xor logic operation, at frequencies reaching up to 50 GHz.

Figure 1

(a) Scheme of the double-pump and probe experimental setup in Faraday geometry used for measuring magnetization precession and imaging of the domain structure and its switching. Both pumps A and B use their own delay lines, allowing for a precise definition of the time delay (A or B) between a single-pump beam and a probe beam, $\mathrm{\Delta }t$, or between two pump pulses, $\mathrm{\Delta }{t}_{\text{AB}}$. (b) Magnetic domain structure with easy-magnetization-axis orientation in YIG:$\mathrm{Co}$ films and magneto-optical images presenting the initial domain structure and the switched structure after irradiation by a single-pump pulse. By applying an 80-Oe external magnetic field with a time duration of about 1 s along the [1−10] or [110] directions, the initial domain structure containing specific pairs of magnetic states (M₁/M₅ and M₈/M₄) can be selected. After the magnetic field is removed, the magnetic pattern stays stable. Due to the labyrinthlike and difficult-to-distinguish character of the changes in the domain structure, all subsequent images are presented in a difference form created by subtracting images before and after irradiation. Marked magnetization states correspond to states aligned along M₁ || [11−1], M₄ || [1−1−1], M₅ || [11−1], and M₈ || [1−11] directions.

Figure 2

(a) Magneto-optical difference images presenting a double-pulse writing sequence with two identical pump pulses A and B with polarization $E_{\text{A}}$ and $E_{\text{B}}$ along the [100] axes and delayed by $\mathrm{\Delta }{t}_{\text{AB}}$. First image shows switching after excitation with either a single A or B pump pulse. Subsequent images present the obtained switching pattern during double-pump pulse excitation with the marked delay, $\mathrm{\Delta }{t}_{\text{AB}}$. (b) Normalized switched area as a function of delay time, $\mathrm{\Delta }{t}_{\text{AB}}$ (circled black points). Solid red line directing to the saturation regions was fitted using $\cos (2\pi \mathrm{\Delta }{t}_{\text{AB}}/t_{\text{sw}})$, where $t_{\text{sw}}=(178.6\pm 5.2)\text{ps}$ is the half period of magnetization precession (T) [14]. Square blue points represent the amplitude determined from the double-pump traces as the amplitude of magnetization precession. Dashed red line is a guide for the eye. (c) Time-resolved Faraday rotation signal from the only single-pump pulse (red points) and the double-pump and probe pulses for different times $\mathrm{\Delta }{t}_{\text{AB}}$. Both pump pulses were polarized along the same polarization plane along the [100] direction in YIG:$\mathrm{Co}$. Curves are offset vertically. Arrows mark the temporal position of pump pulse B, which is delayed by $\mathrm{\Delta }{t}_{\text{AB}}$.

Figure 3

(a) Differential single-shot time-resolved images of the magnetic pattern obtained at Δt₁= 40 ps and Δt₂= 147 ps. Each subsequent image starts as an initial state for the next pump-pulse switching. Observed change in the image intensity corresponds to the change in the normalized magnetization component. Between every captured time-resolved image, the T_R < 50 ms time required for data acquisition; therefore, the initial state is always fully switched. (b) Time-resolved change of the Faraday rotation proportional to the change in the out-of-plane magnetization component, $\mathrm{\Delta }{M}_z$. Traces were retrieved from the image stack from the region marked by colored squares separately for both black, with M− along the [11−1] direction, and white, M+ along the [1−11] direction, domains.

Figure 4

(a) Magneto-optical difference images presenting a double-pulse writing sequence with two pulses A with $E_{\text{A}}$ || [100] axis and B with $E_{\text{B}}$ || [010] axis. As previously, the first image shows switching after excitation with either a single A or B pump pulse with the size used for normalization. Subsequent image represents the obtained switched pattern depending on $\mathrm{\Delta }{t}_{\text{AB}}$. (b) Normalized switched area as a function of delay time, $\mathrm{\Delta }{t}_{\text{AB}}$ (circled black points). Square blue points represent the amplitude determined from the double-pump traces as the amplitude of magnetization precession. Dashed line represents the effective precession amplitude of both pump pulses. (c) Time-resolved Faraday-rotation signal from the double-pump and probe setup with pump A polarized along the [100] plane and pump B along the [010] plane. Notably, the two pump signals, being in counterphase, interfere completely destructively, Δt_AB= 0 ps.