The conventional approach to flipping electron spins in a semiconductor requires an external alternating field. It seems that the same job can be accomplished without external excitation of any kind.
A mother pushing her son on a swing knows that she must push the swing in synchrony with the swinging motion. Similarly, to flip the spin of an electron in a static magnetic field, we must apply an alternating magnetic field whose frequency matches the wobbling motion of the electron's spin orientation about the static field. Or so we thought. In their report on page 868 of this issue, Frolov et al.1 show that electron spins can be flipped with electric fields, which is unusual in itself, and that the fields are purely static. This sounds as if the mother were pushing the swing in the same direction all the time, but the swing nevertheless rocks back and forth.
Spin flips have been achieved using external electric fields in previous studies, but these fields were always oscillating, such that electrons in the material under study were pulled back and forth periodically. This in turn, by means of a quantum effect known as spin–orbit interaction, creates the alternating magnetic field required to flip the spin: when an electron moves through the material, its spin experiences a magnetic field that is proportional to the electron's velocity (orbit). The orientation of this field is reversed when the electron's direction of motion is reversed. This approach has been used to flip spins in a variety of systems, from two-dimensional gases of electrons2 to 'quantum dots'3. But what would be desirable for electronic applications is the ability to manipulate spins without any external excitation, using for example static electric fields, which are easier to generate on a chip than oscillating fields.
Using a clever and original approach, Frolov et al.1 accomplish just that by making electrons bounce back and forth between the two walls of a micrometre-wide semiconductor channel using solely static electric fields. The spin–orbit-induced magnetic field then alternates with a frequency that corresponds to the electron's bouncing frequency, which depends only on the separation between the channel walls and on the Fermi velocity — the characteristic velocity with which electrons move through the semiconductor.
In the authors' experiment, electrons are injected into the semiconductor channel via a narrow opening that filters electrons, selecting those with a spin axis oriented along a static magnetic field. The electrons then move through the channel, bouncing back and forth between the channel walls along the way. A second opening, 20 micrometres down the channel, acts as a spin detector and thus tests whether the electron spins have been flipped or not (Fig. 1).
But how do Frolov and colleagues know whether their technique of flipping electron spins really works? After all, there can be other reasons — apart from the bouncing between channel walls — for electron spins to flip between injector and detector4. To confirm their technique, the authors performed two tests. The first test consisted of assessing whether a spin flip occurred only when the electron's bouncing frequency matched the spin resonance frequency (the magnetic-field frequency necessary to trigger the spin flip), which is proportional to the strength of the static magnetic field. And indeed, spin flips were detected only for a specific magnetic field strength. Furthermore, the required magnetic-field strength changed as predicted with the channel width, as well as with the Fermi velocity (which is controlled by the electron density), both of which affect the bouncing frequency.
The second crucial test involved aligning the static magnetic field along the channel axis rather than perpendicular to it, as was the case in the first test. The spin–orbit-induced magnetic field was then parallel to the induced field expected, which should not — and did not — lead to spin flips. The two tests thus confirm the authors' novel technique as a robust method of flipping spins.
But there's a catch. As far as electronic applications are concerned, the technique comes with a major limitation: it does not flip electron spins in a coherent manner, but rather does so randomly — that is, it cannot rotate the electron spins to a specific orientation. Although the authors used an ultraclean semiconductor material, electrons still scatter off charged impurities in the vicinity of the channel, randomizing the bouncing path, and hence the rotation of their spins. In addition, electrons enter the semiconductor channel and hit the walls with a range of incidence angles (Fig. 1), leading to a large array of bouncing frequencies and hence of spin orientations. But it may be possible to circumvent this limitation by using even cleaner materials and, as Frolov and colleagues suggest, by applying electron-focusing techniques to obtain a well-defined angle of incidence and a more sharply defined bouncing frequency.
In the future, one could imagine applying the authors' technique of flipping spins to entire solid-state electronic circuits in which information is encoded in the spin state of the electrons. This is the vision of the field of spintronics5, which has already led to discoveries such as 'giant magnetoresistance' and the subsequent miniaturization of hard-disk drives. One step beyond lies the coherent control of quantum-mechanical superpositions of electron spin states6, which may one day lead to applications in quantum-information processing.
Frolov, S. M. et al. Nature 458, 868–871 (2009).
Kato, Y., Myers, R. C., Gossard, A. C. & Awschalom, D. D. Nature 427, 50–53 (2004).
Nowack, K. C., Koppens, F. H. L., Nazarov, Yu. V. & Vandersypen, L. M. K. Science 318, 1430–1433 (2007).
Meier, F. & Zakharchenya, B. P. (eds) Optical Orientation (North-Holland, 1984).
Wolf, S. A. et al. Science 294, 1488–1495 (2001).
Hanson, R. et al. Rev. Mod. Phys. 79, 1217–1265 (2007).