Current-Induced Spin Polarization in Strained Semiconductors

The polarization of conduction electron spins due to an electrical current is observed in strained nonmagnetic semiconductors using static and time-resolved Faraday rotation. The density, lifetime, and orientation rate of the electrically polarized spins are characterized by a combination of optical and electrical methods. In addition, the dynamics of the current-induced spins are investigated by utilizing electrical pulses generated from a photoconductive switch. These results demonstrate the possibility of a spin source for semiconductor spintronic devices without the use of magnetic materials.

Figure 1

(a) Device schematic and sample geometry. Dark areas are contacts and light gray area is the InGaAs channel. (b) Schematic of experimental setup and Larmor precession of spins excited by electrical current. (c) Voltage-induced ${\theta }_F$ as a function of $B$ for $E=4$, 8, and $12\mathrm{m}\mathrm{V}\mu {\mathrm{m}}^{-1}$ ($\mathbf{E}//[1\overline{1}0]$), taken at $T=5\mathrm{K}$ from a device with $w=60\mu \mathrm{m}$ and $l=200\mu \mathrm{m}$. Open circles are data, and lines are fits as explained in the text. Constant offsets have been subtracted for clarity.

Figure 2

(a) Time-resolved Faraday rotation data in the absence of electric fields at $T=5\mathrm{K}$ and $B=0.00\mathrm{T}$ showing ${\theta }_{\mathrm{o}\mathrm{p}}$ (filled square) due to circularly polarized pump pulse with $P=2.2\mu \mathrm{W}$. Open circles are data, the dotted line shows the extrapolation used for obtaining ${\theta }_{\mathrm{o}\mathrm{p}}$, and the solid line is a guide to the eye. Inset shows the spatial scans of ${\theta }_F$ taken at $E=0\mathrm{m}\mathrm{V}\mu {\mathrm{m}}^{-1}$, $T=5\mathrm{K}$ and $B=0.00\mathrm{T}$. $\Delta t=50\mathrm{p}\mathrm{s}$ for the scan along $x$ (open circles) and $\Delta t=10\mathrm{p}\mathrm{s}$ for the scan along $y$ (open squares). Lines are Gaussian fits. (b), (c) , and (d) show the spin density ${\rho }_{\mathrm{e}\mathrm{l}}$, the spin lifetime $\tau$, and the spin orientation rate $\gamma$, respectively, as a function of $E$ for $\mathbf{E}//[1\overline{1}0]$ (filled circles, taken from a device with $w=60\mu \mathrm{m}$ and $l=200\mu \mathrm{m}$) and $\mathbf{E}//[1\overline{1}0]$ (open circles, taken from a device with $w=80\mu \mathrm{m}$ and $l=300\mu \mathrm{m}$). Overall scaling of ${\rho }_{\mathrm{e}\mathrm{l}}$ and $\gamma$ has errors up to 40%.

Figure 3

Temperature dependence of ${\rho }_{\mathrm{e}\mathrm{l}}$ (a), $\tau$ (b), $\gamma$ (c), and $\eta$ (d). $\mathbf{E}//[1\overline{1}0]$ and data taken from a device with $w=60\mu \mathrm{m}$ and $l=200\mu \mathrm{m}$.

Figure 4

(a) Device schematic. (b) Time evolution of voltage-induced ${\theta }_F$. Top curves (red, $B=+44\mathrm{m}\mathrm{T}$, blue, $B=-44\mathrm{m}\mathrm{T}$) show the raw data. Black curve is the background signal. Bottom curves show the data after background subtraction. (c) ${\theta }_F$ (background subtracted) as a function of $\Delta t$ and $B$. (d) and (e) show ${\omega }_L$ and $\varphi$, respectively, obtained from fits to data in (c).