Measuring spin-injection and detection in ferromagnet/semiconductor junctions
Over the past few years a great deal of attention has been paid to the field of spintronics, in which both the spin and the charge of electrons are used for devices. The first spintronic devices were the all metallic GMR read heads which appeared in the late nineties. However, many proposed spintronic devices involve semiconductors due to the long spin lifetimes and the possibilities for manipulating spins. The ability to control the electron spins allows spintronic devices to be used for logical and quantum computing applications as well as the conventional memory applications. A prerequisite for the realisation of these new concepts is that of achieving both efficient spin-injection from a ferromagnet into a semiconductor and spin-detection of a polarised current passing from a semiconductor into a ferromagnet. In the TFM group we examine both these processes by either optically exciting or detecting spin-polarized electrons in the semiconductor.
Figure 1: Schematic of the spin FET as proposed by Datta and Das. The dotted white line denotes the 2DEG formed at the InGaAs/InAlAs interface. The numbers (I?III) indicate the important spin dependent processes, I- spin injection, II- manipulation of spins and III- spin detection.
To date, two different experimental approaches for spin injection into semiconductors have been reported, using either diluted magnetic semiconductors or ferromagnetic metals as a spin aligner. While high spin injection efficiencies (up to about 90%) have been reported for magnetic semiconductor electrodes at low temperatures, their implementation in spintronic devices (which need to be operated at room temperature) is difficult due to their low Curie temperature TC. There has been extensive effort to increase the TC of magnetic semiconductors with some reports of room temperature ferromagnetism. In order to achieve a large spin polarisation, however, TC has to be considerably larger than room temperature and this remains a challenging goal in current research. Ferromagnetic metals such as Fe, Co and Ni, on the other hand, have a high enough TC for room temperature operation and low saturation fields, as required for spintronic applications. The focus of the current research in the TFM group lies on the investigation of spin transport across ferromagnetic metal/semiconductor interfaces.
Fortunately, we can measure the properties of the spin-injection and spin-detection processes separately because of the optical selection rules for zincblende semiconductors, these are shown below in figure 2. This allows us to investigate the spin-injection process by measuring the circularly polarized light emitted from a ferromagnet/semiconductor diode (we use Fe on GaAs mainly). We can investigate the spin-detection process by measuring the current across a similar diode made from Fe grown on a GaAs substrate when its is illuminated by circularly polarized light. More details of the experiments are discussed below.
Figure 2:The optical selection rules for zincblende semiconductors (e.g. GaAs) a) for when the light and heavy hole states are degenerate (e.g. in the bulk) and (b) for when they are non-degenerate (e.g. in a quantum well or when strained), these allow the spin polarization of any current in the semiconductor to be measured. The transitions shown are for radiative recombination, but the rules also work (in reverse) for photoexcitation. Note the light involved is circularly polarized.
The sample, shown in figure 3, is a GaAs / AlGaAs quantum well heterostructure with an Fe (magnetic) or Au (non-magnetic) metallic layer to act as the top "spin-injector" electrode. We work in reverse bias so electrons are injected from the metallic layer into the heterostructure. We use a quantum well as the spin-detector to enhance the radiative recombination efficiency, although this has a pay off with a greater spin relaxation rate.
We detect the light, which is in the near infrared, through the top contact this means that MCD may be present in our results. To measure the circular polarization of the emitted light we use a quarter waveplate followed by a linear polarizer, energy resolution is obtained using a monochromator. The degree of circular polarization is given by Pcirc=I(B)-I(-B)/I(B)+I(-B), this can be converted into a spin polarization using the selection rules (figure 2) and so Pspin = Pcirc for our quantum well samples, and Pspin=2Pcirc for bulk like samples.
Figure 3: a) schematic of the semiconductor heterostructure used, note the buried quantum well and that a Schottky barrier will be formed at the Fe/GaAs interface. b) a photograph of the experimental apparatus.
Figure 4:This shows the degree of circular polarization measured against the wavelength. The red stars are from a magnetic Fe/GaAs sample whereas the blue squares are from a non-magnetic Au/GaAs sample. The features in the Fe data, thus have a clear magnetic origin and demonstrate spin injection into the GaAs even if the efficiency is low.
The aim of this project is to investigate spin dependent electron transport across GaAs/ferromagnetic (FM) metal interfaces. In our experiments, spin polarised electrons are optically created in the GaAs by illumination with circularly polarised light. The photoexcited electrons passing from the semiconductor (SC) into the FM have different transmission probabilities at the SC/FM interface depending on their spin orientation with respect to the FM layer magnetisation. In this way, a spin imbalance of the electron current can be detected and measured as a modulation of the photocurrent.
Depending on the applied bias, different charge transport processes (thermionic emission, quantum-mechanical tunnelling, ballistic transport) will contribute to the net photocurrent in a FM/GaAs structure, as is schematically shown in figure 5 for the case of a direct Schottky interface. In order to study in detail the role of these transport mechanisms for spin detection, we investigated (NiFe, Fe)/GaAs Schottky [1,2] and artificial tunnel barrier [3,4] structures at various temperatures
Figure 5:Schematic of (a) the photoexcitation set up and (b) the photoexcited charge transport processes in a FM/SC Schottky barrier structure at moderate forward bias: thermionic electron emission (ITE), electron tunnelling (IT), electron and hole diffusion (ID). The arrow thickness denotes the probability of the different processes
Figure 6:FM layer magnetisation (solid lines, measured with MOKE) and spin detection efficiency (solid symbols) vs. magnetic field for two NiFe/GaAs (ND = 1023 m-3) samples with different NiFe thicknesses. In the inset, a schematic of the electron tunnelling process in a FM/SC Schottky barrier structure is shown. The arrow thickness denotes the transmission probability across the interface. Up-spin and down-spin electrons have the same excitation energy but are shifted along the energy axis here for a better illustration of the transport mechanism.
With regard to spintronic applications, it will be the major aim of future experiments to further improve the spin detection efficiency. This could be achieved by:
- increasing the photoexcited electron spin polarisation (to up to 100%) by using strained GaAs substrates or quantum well structures
- using highly spin polarised FM electrodes, such as for example Heusler alloys
- optimising the FM/SC interface properties and the tunnel barrier characteristics
- replacing the single FM layer with a more complicated metal multilayer structure, such as a spin valve; spin detection efficiencies of up to 60% have recently been observed in our group for Co/Cu/NiFe/GaAs structures 
1. J. A. C. Bland, S. J. Steinmuller, A. Hirohata, and T. Taniyama, Optical Studies of Electron Spin Transmission, in J. A. C. Bland and B. Heinrich, editors, Ultrathin Magnetic Structures IV: Applications of Nanomagnetism, Springer Verlag, Berlin (2005)
2. S. J. Steinmuller, C. M. Guertler, G. Wastlbauer and J. A. C. Bland, Phys. Rev. B, in press
3. T. Taniyama, G. Wastlbauer, A. Ionescu, M. Tselepi, and J. A. C. Bland, Phys. Rev. B 68, 134430 (2003)
4. S. E. Andresen, S. J. Steinmuller, A. Ionescu, G. Wastlbauer, C. M. Guertler and J. A. C. Bland, Phys. Rev. B 68, 073303 (2003)
5. S. J. Steinmuller, T. Trypiniotis, W. S. Cho, A. Hirohata, W. S. Lew, C. A. F. Vaz and J. A. C. Bland, Phys. Rev. B 69, 153309 (2004)