Spin transfer experiments on $(\mathrm{Ga},\mathrm{Mn})\mathrm{As}∕(\mathrm{In},\mathrm{Ga})\mathrm{As}∕(\mathrm{Ga},\mathrm{Mn})\mathrm{As}$ tunnel junctions

We present and discuss the results of current induced magnetic switching experiments performed on pillar-shaped $(\mathrm{Ga},\mathrm{Mn})\mathrm{As}∕(\mathrm{In},\mathrm{Ga})\mathrm{As}∕(\mathrm{Ga},\mathrm{Mn})\mathrm{As}$ tunnel junctions. The sign of the switching currents confirms the opposite spin polarizations of the valence band holes and Mn atoms in (Ga,Mn)As. With respect to spin transfer experiments in purely metallic structures, the magnitude of the switching currents is smaller by two orders of magnitude, which can be explained mainly by the small magnetization of (Ga,Mn)As. A striking result is the observation of current induced magnetization switching at values of the bias voltage for which the magnetoresistance of the junction has dropped to almost zero. This raises interesting questions on the different role played by voltage-induced magnon excitations on magnetoresistance and current-induced magnetization switching.

Figure 1

(Color online) Tunnel magnetoresistance of a $(\mathrm{Ga},\mathrm{Mn})\mathrm{As}∕(\mathrm{In},\mathrm{Ga})\mathrm{As}∕(\mathrm{Ga},\mathrm{Mn})\mathrm{As}$ tunnel junction $(\text{diameter}=700\mathrm{nm})$ recorded at $3\mathrm{K}$ and $1\mathrm{mV}$ with the field along [100] (the first magnetic reversal corresponds to the thick magnetic layer). Inset: $I\left(V\right)$ curve at $3\mathrm{K}$ in the parallel configuration.

Figure 2

Tunnel magnetoresistance versus bias voltage (bottom scale) and current density (top scale) at $3\mathrm{K}$.

Figure 3

(Color online) (a) Resistance versus magnetic field at $3\mathrm{K}$ and $20\mathrm{mV}$. (b) Resistance versus current density at $3\mathrm{K}$ (initial temperature at low current density) in a field of $13.5\mathrm{Oe}$ (the initial state has parallel magnetizations in the positive field direction). (c) Same as (a) but at $30\mathrm{K}$. (d) Same as (b) but at $30\mathrm{K}$.

Figure 4

(Color online) Coercive field of the thick (dots) and thin layers (squares) (a) versus temperature from MR cycles at low bias and (b) versus voltage from MR cycles under several bias at $3\mathrm{K}$ (initial temperature at low bias). The coercive field is corrected for the dipole field acting on each layer, that is $H_c=H_c{}^{\mathit{exp}}-H_{\mathit{dip}}$ (the dipole field $H_{\mathit{dip}}$, acting on each layer, is derived from minor MR cycles).