The operation of traditional semiconductor devices depends on the transport and storage of electronic charge. But electrons have spin as well as charge, and taking advantage of the spin could revolutionize electronics, leading to new devices such as spin transistors1, spin memory storage or even spin quantum computers2. One requirement for such devices is the efficient injection of spin-polarized carriers (electrons, or their positively charged counterparts, holes) into a semiconductor. This has proved to be a huge challenge. Now two groups have independently achieved success with real electrical devices. On pages 787 and 790 of this issue, Fiederling et al.3 and Ohno et al.4 report the injection of spin-polarized electrons or holes into a light-emitting diode, with efficiencies of about 90% and 2%, respectively. In both cases the diodes emit circularly polarized light, confirming the spin polarization of the carriers. So far these devices only work at low temperatures, but the race towards commercial semiconductor spin electronics is on.
Why is spin electronics in semiconductors interesting? One motivation comes from the discovery that alternating layers of non-magnetic materials and ferromagnetic metals (such as iron) show large changes in their electrical resistance when a small magnetic field is applied. This change, known as giant magnetoresistance, can be greater than 60% at room temperature and results from the specific electron-spin orientation across the layers5,6. The effect is already exploited in many applications, such as in the millions of reader heads in high-density magnetic computer disks. Another example, which might have a huge economic impact in the future, is nonvolatile magnetic random access memory that can retain information when switched off7. The use of spin effects in metals has led rapidly to commercial applications. Even more enticing is the dream of spin electronics in semiconductors, where you can easily engineer the electronic properties, combine conventional and spin electronics on one chip, and build opto-electronic spin devices.
Spin injection is currently the biggest obstacle facing semiconductor spin electronics. Why is this? Many groups have tried to use ferromagnetic metal contacts to inject spin-polarized electrons directly into a non-magnetic semiconductor. This seems like a good idea because electrons in ferromagnetic metals already have a preferential spin orientation at room temperature, even without an external magnetic field. But so far experiments have shown little or no spin polarization of carriers injected from a ferromagnetic metal into a semiconductor8. Dead magnetic layers at the metal–semiconductor interface and the huge difference in the number and energy of carriers are possible causes.
At the interface between a magnetic and a non-magnetic semiconductor these problems are absent. This is good news because, as yet, spin electronics has produced all-metal devices or devices combining metals and semiconductors. The goal is to make an all-semiconductor device, which would make it easier to incorporate spin electronics with traditional semiconductor technology. The two experiments reported in this issue3,4 show clearly that unpolarized carriers can become strongly spin polarized in a magnetic semiconductor and then electrically injected (that is, voltage driven) into a non-magnetic semiconductor with their spin orientation intact. Is the problem of spin injection in semiconductors finally solved?
Fiederling et al.3 show that voltage-driven spin injection from a semimagnetic to a non-magnetic semiconductor is highly efficient: the device injects 90% spin-polarized current into a light-emitting diode, which is good enough for industrial spin devices. They use an unusual semimagnetic semiconductor (a group II–VI compound; BeMnZnSe) as the ‘spin aligner’ to polarize the electrons before they are injected into the non-magnetic semiconductor (GaAs). Because the GaAs device is configured as a light-emitting diode, the polarized current leads to polarized light emission. The spin aligner works well at low temperatures and provides a new technique for probing spin transport in semiconductors. But it has been known for some time that this material is only semimagnetic at temperatures much below liquid nitrogen (a few kelvin). It is probably not possible to improve this temperature range by increasing the magnetic Mn concentration, because Mn ions usually couple antiferromagnetically in this system at high concentrations. A ferromagnetic II–VI semiconductor working as a spin aligner at room temperature would therefore be a great surprise.
Ohno et al.4 take a different approach. They use GaMnAs (a group III–V semiconductor) as a spin aligner for positively charged holes (rather than electrons). Spin-polarized holes are electrically injected from the GaMnAs, through a GaAs spacer, into a quantum well configured as a light-emitting device. At the same time unpolarized electrons are injected into the quantum well, where they recombine with the holes, emitting polarized light. GaMnAs has a huge advantage over II–VI semiconductors because it can be ferromagnetic at moderate temperatures and perhaps even at room temperature if the Mn concentration can be increased. Moreover, spin injection in such ferromagnetic semiconductors does not require an external magnetic field.
The polarization of the holes in the GaMnAs spin aligner is probably close to 100%, similar to the electron spin polarization seen by Fiederling et al.. But the polarization of the emitted light, which is used to probe the spin of the injected carrier, is more than ten times lower in Ohno and colleagues' device. The reasons are not yet clear but two explanations should be considered. First, the optical selection rules for heavy-hole transitions in ideal quantum wells prohibit circularly polarized light emission in certain directions (normal to the growth direction), even if all the hole spins are polarized. The correlation between carrier spin and polarization of the emitted light is therefore not as straightforward as in Fiederling and colleagues' device. Second, the orientation of the hole spin in bulk GaAs is, unlike the orientation of the electron spin, extremely unstable, and even in quantum wells the hole-spin orientation disappears incredibly quickly at moderate hole temperatures9,10. Therefore, hole-spin injection will probably not be important for future spin electronics.
Despite these limitations, the two experiments represent real progress in spin electronics and they directly point the way to commercial spin-sensitive semiconductor devices. Any ferromagnetic semiconductor that injects electrons and can operate at room temperature will solve the problem, and so the search goes on.
Datta, S. & Das, B. Appl. Phys. Lett. 56, 665–667 (1990).
Imamoglu, A. et al. Phys. Rev. Lett. 83, 4204–4207 (1999).
Fiederling, R. et al. Nature 402, 787–790 (1999).
Ohno, Y. et al. Nature 402, 790–792 (1999).
Grünberg, P., Schreiber, R., Pang, Y., Brodsky, M. B. & Sowers, H. Phys. Rev. Lett. 57, 2442–2445 (1986).
Schad, R. et al. Phys. Rev. B 59, 1242–1248 (1999).
Parkin, S. S. P. et al. J. Appl. Phys. 85, 5828–5833 (1999).
Hammar, P. R., Bennet, B. R., Yang, M. J. & Johnson, M. Phys. Rev. Lett. 83, 203–206 (1999).
Meier, F. & Zakharchenya, B. P. (eds),. Optical Orientation (North-Holland, Amsterdam, 1984).
Marie, X. et al. Phys. Rev. B 60, 5811–5817 (1999).
About this article
Ferromagnetic resonance and control of magnetic anisotropy by epitaxial strain in the ferromagnetic semiconductor (Ga0.8,Fe0.2)Sb at room temperature
Physical Review B (2019)
Journal of Physics and Chemistry of Solids (2017)
Impact of interfacial scattering on the spin polarization of a metal/semiconductor/metal with Rashba spin–orbit coupling junction
Physica B: Condensed Matter (2015)
Journal of the American Chemical Society (2015)
Materials Research Bulletin (2014)