Stochastic Current-Induced Magnetization Switching in a Single Semiconducting Ferromagnetic Layer

We show experimental evidence of magnetization switching in a single (Ga,Mn)(As,P) semiconducting ferromagnetic layer, attributed to a strong reduction of the magnetization and the anisotropy due to current injection. The nucleation of magnetization reversal is found to occur even in the absence of a magnetic field and to be both anisotropic and stochastic. Our findings highlight a new mechanism of magnetization manipulation based on spin accumulation in a semiconductor material.

Figure 1

Experimental set-up and stochastic magnetization reversal. (a) Optical micrograph of the (Ga,Mn)(As,P) layer and schematic of the electrical connections showing the semicircular (left) and narrow (right) $\mathrm{Au}/\mathrm{Ti}$ electrodes. (b) Cross section of the narrow electrode featuring the current flow, the volume ($V_{\mathrm{SF}}\approx w{l}_{\mathrm{SF}}^2$, see text) where hole spins accumulate (semicircle not to scale) and the decrease of the magnetization (arrows). (c) Magnetic domains observed by polar magneto-optical Kerr microscopy. The two gray levels (up/down) correspond to magnetization pointing towards the $+z$ and $-z$ directions. The sample is initially in a saturated magnetic state (homogeneous bright gray level). The injected current pulses with an intensity of $I=2.164\mathrm{mA}$ and durations of $1\mu \mathrm{s}$, $10\mu \mathrm{s}$ and $100\mu \mathrm{s}$ produce a set of quasi-circular domains revealing successive magnetization switching. Images were acquired $30\mathrm{s}$ after the current pulses were injected. Initial sample temperature $T_0=95\mathrm{K}$. No magnetic field is applied. (d) Inset. Typical random telegraph noise reflecting variations of the gray level in a $100\mu {\mathrm{m}}^2$ square close to the interface. Semi-log plot of the typical probability of non-reversal (${\overline{P}}_{\text{up}}$ and ${\overline{P}}_{\text{down}}$) as a function of the waited time ($t$) since the last reversal. Each curve is deduced from $\approx 50$ switches. The (green) lines are exponential fits.

Figure 2

Variation of the dwell times with the external parameters (in semilogarithmic scales). (a) Magnetic field perpendicular to the film ${\mu }_0{H}_z$. (b) Bias current $I$. Interface temperatures and cryostat temperatures ($T_0$) are indicated on top of the curves. The solid lines are fits of Eq. (3). (c) Magnetic field parallel to the film ${\mu }_0{H}_{xy}$. $\theta$ is the angle between ${\mu }_0{H}_{xy}$ and the [100] crystallographic direction.

Figure 3

In-plane magnetic field contributions to the dwell times. (a) ${\tau }_{\text{diff}}$, in semilogarithmic scale and (b) ${\tau }_{\text{mean}}$ (see text). In (a), the solid lines are guides for eyes. In (b), the solid lines are fits based on Eq. (3) for $T=98\mathrm{K}$, $I=0.85\mathrm{mA}$ and for $T=92\mathrm{K}$, $I=2.06\mathrm{mA}$. The angles between the applied magnetic field and the [110] crystallographic direction are indicated at the bottom right of each graph.

Figure 4

Ferromagnet magnetic states close to the interface. In a simplified way, three areas can be distinguished: (1) for $0<l<l_{\mathrm{SF}}$, the spin accumulation produces a strong out-of-equilibrium Mn-carrier spins system; (2) for $l_{\mathrm{SF}}<l<l_{\text{relax}}$, the Mn and carrier spins systems are coupled and the magnetization relaxes towards equilibrium; and (3) for $l>l_{\text{relax}}$, the magnetization is at equilibrium.