Figure 1

Resistivity vs temperature measurement of one of the constrictions whose ac resistance measurements are presented in the paper. Inset: SEM image of a $200\mathrm{nm}$ wide constriction, patterned by FIB.Reuse & Permissions

Figure 2

(a) Differential resistance $dV∕dI$ characteristics as a function of the superimposed dc bias current (which is cycled from zero, to its maximum positive value, then to its maximum negative value, then again to zero) measured at different magnetic fields and at temperature $10\mathrm{K}$ in a typical constriction. The black arrows roughly indicate the threshold current for the onset of hysteretic features. The curves are shifted by a constant offset for clarity. (b) Differential resistance $dV∕dI$ characteristics as a function of the superimposed dc bias current measured in the same constriction in zero magnetic field at $10\mathrm{K}$, by sweeping either the positive current branch first (open circles) or the negative current branch first (filled squares); the numbers in the circles/squares from 1 to 4 and the black arrows indicate the temporal succession of the branches and the direction of the current sweep for each branch. (c) Differential resistance $dV∕dI$ characteristics as a function of the superimposed dc bias current measured in the same constriction in zero magnetic field at different temperatures; the black arrows roughly indicate the threshold current for the onset of hysteretic features. The curves are shifted by a constant offset for clarity.Reuse & Permissions

Figure 3

Differential resistance $dV∕dI$ characteristics as a function of the superimposed dc bias current (only the one-way sweep is shown) measured at different magnetic fields and at temperature $10\mathrm{K}$ in another typical constriction. The shift of the features at higher currents with increasing magnetic field is clearly visible.Reuse & Permissions

Figure 4

Differential resistance $dV∕dI$ characteristics as a function of the superimposed dc bias current (which is cycled from zero, to its maximum positive value, then to its maximum negative value, then again to zero as indicated by the black arrows) measured at different magnetic fields and at temperature $10\mathrm{K}$ in an uncostricted mesa. The curves are shifted by a constant offset for clarity.Reuse & Permissions

Figure 5

Schematic picture of adjacent magnetic domains in four different initial configurations. The polarized electron flow is assumed from right to left (in the opposite case all the signs of the torques should be reversed). The thick arrows ${\mathbf{s}}_{\mathbf{A}}$, ${\mathbf{s}}_{\mathbf{B}}$, and ${\mathbf{s}}_{\mathbf{C}}$ indicate the direction of each domain magnetization. The larger domains $A$ and $C$ have larger coercive fields than the smaller domain $B$, so that only the latter rotates, due to the spin torques ${\mathbf{\tau }}_{\mathbf{A}}$ and ${\mathbf{\tau }}_{\mathbf{C}}$. The torque ${\mathbf{\tau }}_{\mathbf{A}}$ is due to the electrons tunneling from the domain $A$ to the domain $B$, while the torque ${\mathbf{\tau }}_{\mathbf{C}}$ is due to the electrons tunneling from the domain $B$ to the domain $C$. The directions and signs of ${\mathbf{\tau }}_{\mathbf{A}}$ and ${\mathbf{\tau }}_{\mathbf{C}}$ are determined by the direction of the charge-carrier flow and by the direction of the local magnetization of the adjacent domains, according to the theory of spin torque.Reuse & Permissions

Figure 6

(a) dc current-voltage characteristics of a constriction measured at different magnetic fields at $10\mathrm{K}$. (b) Constriction differential resistance obtained from the curves of the upper panel (a) by numerical differentiation. The ellipses evidence the sharp changes of slope observed for negative sign of the applied current.Reuse & Permissions