A. T. Filip

Fundamental obstacle for electrical spin injection from a ferromagnetic metal into a diffusive semiconductor

We have calculated the spin-polarization effects of a current in a two-dimensional electron gas which is contacted by two ferromagnetic metals. In the purely diffusive regime, the current may indeed be spin-polarized. However, for a typical device geometry the degree of spin-polarization of the current is limited to less than 0.1% only. The change in device resistance for parallel and antiparallel magnetization of the contacts is up to quadratically smaller, and will thus be difficult to detect.

Electrical spin injection and accumulation at room temperature in an all-metal mesoscopic spin valve.

Finding a means to generate, control and use spin-polarized currents represents an important challenge for spin-based electronics, or ‘spintronics’. Spin currents and the associated phenomenon of spin accumulation can be realized by driving a current from a ferromagnetic electrode into a non-magnetic metal or semiconductor. This was first demonstrated over 15 years ago in a spin injection experiment on a single crystal aluminium bar at temperatures below 77 K. Recent experiments have demonstrated successful optical detection of spin injection in semiconductors, using either optical injection by circularly polarized light or electrical injection from a magnetic semiconductor. However, it has not been possible to achieve fully electrical spin injection and detection at room temperature. Here we report room-temperature electrical injection and detection of spin currents and observe spin accumulation in an all-metal lateral mesoscopic spin valve, where ferromagnetic electrodes are used to drive a spin-polarized current into crossed copper strips. We anticipate that larger signals should be obtainable by optimizing the choice of materials and device geometry.

Electrical detection of spin precession in a metallic mesoscopic spin valve

To study and control the behaviour of the spins of electrons that are moving through a metal or semiconductor is an outstanding challenge in the field of ‘spintronics’, where possibilities for new electronic applications based on the spin degree of freedom are currently being explored. Recently, electrical control of spin coherence and coherent spin precession during transport was studied by optical techniques in semiconductors. Here we report controlled spin precession of electrically injected and detected electrons in a diffusive metallic conductor, using tunnel barriers in combination with metallic ferromagnetic electrodes as spin injector and detector. The output voltage of our device is sensitive to the spin degree of freedom only, and its sign can be switched from positive to negative, depending on the relative magnetization of the ferromagnetic electrodes. We show that the spin direction can be controlled by inducing a coherent spin precession caused by an applied perpendicular magnetic field. By inducing an average precession angle of 180°, we are able to reverse the sign of the output voltage.

Spin Injection and Spin Accumulation in Permalloy–Copper Mesoscopic Spin Valves

We study the electrical injection and detection of spin currents in a lateral spin valve device, using permalloy (Py) as ferromagnetic injecting and detecting electrodes and copper (Cu) as nonmagnetic metal. Our multiterminal geometry allows us to experimentally distinguish different magnetoresistance signals, being (1) the spin valve effect, (2) the anomalous magnetoresistance (AMR) effect, and (3) Hall effects. We find that the AMR contribution of the Py contacts can be much larger than the amplitude of the spin valve effect, making it impossible to observe the spin valve effect in a “conventional” measurement geometry. However, these “contact” magnetoresistance signals can be used to monitor the magnetization reversal process, of the spin injecting and detecting Py contacts. In a “nonlocal” spin valve measurement we are able to completely isolate the spin valve signal and observe clear spin accumulation signals at T = 4.2 K as well as at room temperature. We obtain spin diffusion lengths in Cu of 1 μm and 350 nm at T = 4.2 K and room temperature respectively.

Spin Injection and Transport in Micro- and Nanoscale Devices

Experiments to explore the transfer of a spin-polarized electric current within small devices have been ongoing for nearly 30 years. But attaining the same level of exquisite control over the transport of spin in micro- or nanoscale devices, as currently exists for the flow of charge in conventional electronic devices, remains elusive. Much has been learned since the time of the first demonstrations of spin polarized tunneling by Tedrow and Meservey. During this period we have witnessed the transformation of spin-based electronic devices from laboratory experiments to the realm of commercially available products. This has been driven especially, just in this past decade, by the robust phenomena of giant magnetoresistance (GMR) [1]. Even more recently, magnetic tunnel junction devices, involving transport of spin polarized electrons across interfaces, have proceeded to commercial development [2]. Meanwhile, spin injection devices — and by “injection” we here denote transferal of spin-polarized carriers into an otherwise nonmagnetic conductor (or semiconductor) — have not reached a similar, commercially viable, state of maturation. In fact, it is fair to say that, at present, even the fundamental physics and materials science of the spin injection process remains in need of significant elucidation.

Direct experimental comparison between the electrical spin injection in a semiconductor and in a metal

Abstract The electrical injection of spin polarized electrons in a semiconductor can be achieved in principle by driving a current from a ferromagnetic metal, where current is known to be significantly spin polarized, into the semiconductor or normal metal via ohmic conduction. For detection a second ferromagnet can be used as drain. In this paper we address the issue of the efficiency of such an approach to spin injection. For this purpose, we made submicron multiterminal lateral spin valve junctions, with NiFe ferromagnetic electrodes. The ferromagnets were making good ohmic contact either to a two-dimensional electron gas (2DEG) channel, or to a Cu channel. In the all-metal case we observe a clear spin accumulation signal. Due to spurious magnetoresistive contribution of the ferromagnetic electrodes, this could only be detected in a non-local geometry. Despite all our efforts, we have not been able to observe spin injection in the semiconductor. We show that both results are in quantitative agreement with the theoretical predictions based on conductivity mismatch arguments.