A novel spin-spin coupling mechanism that occurs during the transport of spin-polarized minority electrons in semiconductors is described. Unlike the Coulomb spin drag, this coupling arises from the ambipolar electric field which is created by the di
fferential movement of the photoelectrons and the photoholes. Like the Coulomb spin drag, it is a pure spin coupling that does not affect charge diffusion. Experimentally, the coupling is studied in $p^+$ GaAs using polarized microluminescence. The coupling manifests itself as an excitation power dependent reduction in the spin polarization at the excitation spot textit{without} any change of the spatially averaged spin polarization.
The effect of an electric field on the spatial charge and spin profiles of photoelectrons in p+-GaAs is studied as a function of lattice and electron temperature. The charge and spin mobilities of photoelectrons are equal in all conditions and exhibi
t the well known increase as the temperature is lowered. It is shown that this is related mainly to the electron statistics rather than the majority hole statistics. This finding suggests that current theoretical models based on degeneracy of majority carriers cannot fully explain the observed temperature dependence of minority carrier mobility.
Spin-polarized transport of photo-electrons in bulk, p-type GaAs is investigated in the Pauli blockade regime. In contrast to usual spin diffusion processes in which the spin polarization decreases with distance traveled due to spin relaxation, image
s of the polarized photo-luminescence reveal a spin-filter effect in which the spin polarization increases during transport over the first 2 microns from 26 % to 38 %. This is shown to be a direct consequence of the Pauli Principle and the associated quantum degeneracy pressure which results in a spin-dependent increase in the minority carrier diffusion constants and mobilities. The central role played by the quantum degeneracy pressure is confirmed via the observation of a spin-dependent increase in the photo-electron volume and a spin-charge coupling description of this is presented.
In p+ GaAs thin films, the effect of photoelectron degeneracy on spin transport is investigated theoretically and experimentally by imaging the spin polarization profile as a function of distance from a tightly-focussed light excitation spot. Under d
egeneracy of the electron gas (high concentration, low temperature), a dip at the center of the polarization profile appears with a polarization maximum at a distance of about $2 ; mu m$ from the center. This counterintuitive result reveals that photoelectron diffusion depends on spin, as a direct consequence of the Pauli principle. This causes a concentration dependence of the spin stiffness while the spin dependence of the mobility is found to be weak in doped material. The various effects which can modify spin transport in a degenerate electron gas under local laser excitation are considered. A comparison of the data with a numerical solution of the coupled diffusion equations reveals that ambipolar coupling with holes increases the steady-state photo-electron density at the excitation spot and therefore the amplitude of the degeneracy-induced polarization dip. Thermoelectric currrents are predicted to depend on spin under degeneracy (spin Soret currents), but these currents are negligible except at very high excitation power where they play a relatively small role. Coulomb spin drag and bandgap renormalization are negligible due to electrostatic screening by the hole gas.
A self-consistent expression for the surface recombination velocity $S$ and the surface Fermi level unpinning energy as a function of light excitation power ($P$) is presented for n- and p-type semiconductors doped above the 10$^{16}$ cm$^{-3}$ range
. Measurements of $S$ on p-type GaAs films using a novel polarized microluminescence technique are used to illustrate two limiting cases of the model. For a naturally oxidized surface $S$ is described by a power law in $P$ whereas for a passivated surface $S^{-1}$ varies logarithmically with $P$. Furthermore, the variation in $S$ with surface state density and bulk doping level is found to be the result of Fermi level unpinning rather than a change in the intrinsic surface recombination velocity. It is concluded that $S$ depends on $P$ throughout the experimentally accessible range of excitation powers and therefore that no instrinsic value can be determined. Previously reported values of $S$ on a range of semiconducting materials are thus only valid for a specific excitation power.
