Research: Spin Transport

Text only
Images

As the name implies, the Spin Transport project investigates the electrical transport of spin-polarized carriers between two materials. Spin transport has been accomplished in several different heterostructures including superconductor-ferromagnet, nonmagnetic metal - ferromagnet or all semiconductor systems in which one of the semiconducting layers is doped with magnetic elements. A natural question is whether the same can be performed for ferromagnet-semiconductor heterostructures, and if so, what the influence of various relevant physical parameters such as doping, bias, and temperature will be.

In addition to this fundamental motivation, a thorough understanding of the physics of spin transport in ferromagnet-semiconductor systems could potentially aid the the development of so-called “spintronic” devices that would exploit the spin of carriers for information processing and data storage. Current semiconductor technologies are based on the transport and storage of charged carriers, and performance improvements are primarily based on the reduction of circuit component size. This approach is ultimately limited by the size of a single molecule, and further gains will eventually require an entirely new technology. Spintronics is one candidate for this new technology.

The important property of quantum variables, such as spin, is that, starting from two initial states, they can create many new states; this phenomenon is known as the superposition principle. This creates a basis for a fundamentally new way of data processing. Traditional microelectronic devices can be in what are called ON and OFF states, thus emulating binary logic. A physical realization of spintronic devices would most likely be based on the fact that conventional ferromagnets such as iron can serve as a good source of spin-polarized carriers while semiconductors serve as good hosts to spins. An interesting question is whether one can build a device (such as a diode and/or transistor) that would operate in multiple states; that is, a pair of ON and OFF states due to transport (or no transport) of charge, and corresponding ON and OFF states due to transport of spin.

To investigate spin transfer, we use a Ferromagnet-Schottky Spin LED. In this technique, a polarized charge carrier (usually an electron) is injected from the ferromagnet into the semiconductor where it recombines with a hole, supplied  from  the  substrate of  the  heterostructure. Such a process (Fig. 1) results in emission of light (electro-luminescence) which is partially circularly polarized, since quanta of light take polarization information from spin-polarized carriers.

Fig. 1: Band Diagram. Spin-polarized electrons (black circles) tunnel across the energy band gap, recombining with holes (white circles). This produces polarized light.

An example of spectral features from a Fe/Al_0.1Ga_0.9As/GaAs/Al_0.1Ga_0.9As heterostructure is shown on the Fig. 2. Here we rely on the fact that the transport of charged carriers in metal-semiconductor contacts is well understood, and we now add spin-polarization of the carriers to that transport.

Fig. 2: Spectral Features from a Fe/Al_0.1Ga_0.9As/GaAs/Al_0.1Ga_0.9As Heterostructure

Having performed magnetic, magneto-optical, and transport measurements on ferromagnet-semiconductor heterostructures, the results of which have encouraged us as to the feasibility of such a process, we are now focused on finding an optimal set of parameters which enhances the observed effects. Compared to more conventional transport techniques, the above method is a rather simple and convincing way of checking the state of polarization of the injected carrier. Analysis of the electroluminescent spectral features is then focused on quantifying what fraction of the light is circularly polarized, whether such circularly polarized light is a result of the recombination between an injected spin-polarized electron and an unpolarized hole, and whether the magnetic field dependence of such a signal correlates with the magnetic properties of the heterostructures. On the Fig. 3 you can see how a scaled, out-of-plane magnetization curve tracks a polarized signal.

Fig. 3: Magnetization Curve Tracks Polarized Signal

In practice, samples comprised of metal/n-type semiconductor/QW/p-type semiconductor (or metal/n-i(QW)-p, that is, a QW imbedded in the depletion region of a n-i-p diode) are prepared by molecular beam epitaxy (MBE) by a collaborating group at U. of M (Hybird Materials Epitaxy Center, C. Palmstrøm). The samples are processed into transport devices using conventional photolithography semiconductor processing techniques in the university’s Microtechnology Lab (MTL). The devices, which now comprise back to back Schottky and p-n diodes, are mounted in a liquid He (LHe) flow cryostat in order to perform measurements at temperatures between 10 K and 100 K.

If the desired measurement is purely electronic, or only requires small magnetic fields, we use a small flow cryostat that affords easy optical access to the device as well as efficient LHe use and quick turn around time. Many of our experiments, however, require the application of large magnetic fields in excess of 2 T, the approximate field at which thin (~50 Å thick) Fe films become magnetized perpendicular to the plane of the film. For these measurements we use a magneto-optical cryostat that incorporates a superconducting magnetic. Devices are biased using a current supply, and the resulting luminescence from carrier combination at the QW is collected using a CCD camera mounted on a spectrometer. Device luminescence is strongest from just under the ferromagnetic contact, that is, from the top of the device. In order to infer information about the spin-polarization of the injected carriers from the polarization state of the emitted light it is necessary to have the angular momentum of the carrier oriented on the axis of light collection, (hence the requirement of greater than 2 T magnetic fields to magnetize the Fe contact out of plane, and as a result along the axis of light collection).

In addition to this primary field dependent, polarization resolved electroluminescence experiment we perform several other experiments (including photoluminescence with linearly and circularly polarized excitation, bias dependence, geometry dependence, and temperature dependence) that further explore the physics of spin transport in these devices and characterize the optical detection scheme that all spin-LED type experiments use.

Our previous projects related to spin transport in ferromagnet-semiconductor heterostructures may be found here. We have explored transport of photoexcited spin-polarized carriers from semiconductors towards ferromagnetic electrodes.