AVS 49th International Symposium
    Electronic Materials and Devices Monday Sessions
       Session EL+SC+MI-MoA

Invited Paper EL+SC+MI-MoA1
Spin Transport in Ferromagnet-Semiconductor Schottky Diodes

Monday, November 4, 2002, 2:00 pm, Room C-107

Session: Metal-Semiconductor Interfaces
Presenter: P.A. Crowell, University of Minnesota
Authors: P.A. Crowell, University of Minnesota
A.F. Isakovic, University of Minnesota
B.D. Schultz, University of Minnesota
J. Strand, University of Minnesota
C.J. Palmstrom, University of Minnesota
Correspondent: Click to Email
We have completed an investigation of spin injection in semiconductor heterostructures using a Schottky contact between Fe and n-Al@sub 1-x@Ga@sub x@As as an injector and an Al@sub 1-x@Ga@sub x@As/GaAs/Al@sub 1-x@Ga@sub x@As quantum well (QW) as the detector. The injector and detector are combined in a single device in which the QW is placed in the depletion region of a p-n junction. The Schottky contact is @delta@-doped, so that a tunneling current can be obtained under moderate reverse bias.@footnote 1@ The injected electrons recombine in the QW with holes from the p-contact, and the polarization of the resulting electroluminescence (EL) is used to infer the spin state of the recombining carriers. We demonstrate that the doping profile chosen for the QW has a dramatic effect on the apparent spin-detection efficiency. EL polarizations over 10% are obtained in optimally biased devices in which the QW is intentionally p-doped. The field-dependence of the EL polarization closely matches the magnetization of the Fe electrode. However, the largest polarizations are not observed from ordinary ground-state recombination in the quantum well. The maximum polarization observed from ground-state recombination is approximately 4% and appears to be less sensitive to the doping profile. In contrast, the EL polarization in control samples is less than 2%, does not track the magnetization of the ferromagnetic electrode and depends only weakly on bias voltage. The interpretation of these measurements will rely on a thorough understanding of the QW spin detector and the identification of background contributions. For example, we show using optical pumping measurements that the spin detection efficiency of the QW is a function of the bias voltage, as is the background photoluminescence polarization. This work was supported by DARPA, ONR, and NSF (MRSEC). @FootnoteText@ @footnote 1@ H.J. Zhu et al., Phys. Rev. Lett. 87, 016601 (2001); A.T. Hanbicki et al., Appl. Phys. Lett. 80, 1240 (2002).