Abstract
The exchange interaction between electrons and magnetic spins is considerably enhanced near interfaces, in magnetic multilayers. As a result, a dc current can be used to generate spin oscillations. We review theory and experimental evidence. The s–d exchange interaction causes a rapid precession of itinerant conduction-electron spins s around the localized spins S of magnetic electrons. This s precession has been observed directly [Weber et al., Science 291, 1015 (2001)] with electron beams through Fe, Co, and Ni films. Because of it, the time-averaged interaction torque between s and S vanishes. Thus, electrons do not interact at all with long-wavelength spin waves, in the bulk. An interface between a magnetic layer and a spacer causes a local coherence between the precession phases of different electrons, in a region within 10 nm from the interface [J. C. Slonczewski, J. Magn. Magn. Mater. 159, L1 (1996)]; [L. Berger, Phys. Rev. B 54, 9353 (1996)]. Also, a second magnetic layer with pinned S is used to “prepare” s in a specific direction, before the electrons cross over to the active magnetic layer. The current-induced drive torque of s on S in the active layer may be calculated from the spin current or from the spin imbalance Δμ. This torque is equivalent to negative Gilbert damping, leading to instability of S. Spin current and Δμ are proportional to each other [L. Berger, J. Appl. Phys. 89, 5521 (2001)] and can arise from Fermi-surface translation, as well as from expansion/contraction. In fields H normal to layers, the critical current Ic for S instability is predicted to be proportional to the ferromagnetic-resonance frequency ω {consistent with Tsoi et al. [Tsoi et al., Phys. Rev. Lett. 80, 4281 (1998); 81, 493 (1998); Nature (London) 406, 46 (2000)] experiments}. However, for in-plane H, due to elliptic S precession, Ic is not proportional to ω, but linear in H [Katine et al., Phys. Rev. Lett. 84, 3149 (2000), experiments]. Apart from the current-induced drive torque, an extra Gilbert damping is predicted near the interface even at zero current [L. Berger, Phys. Rev. B 54, 9353 (1996)]. It has been observed by ferromagnetic resonance [Urban et al., Phys. Rev. Lett. 87, 217204 (2001)].
All Related Versions