JOURNAL OF APPLIED PHYSICS 105, 07D110 共2009兲
Spin-current-induced dynamics in ferromagnetic nanopillars of lateral spin-valve structures J.-B. Laloë,1,a兲 T. Yang,1 T. Kimura,1,2 and Y. Otani1,2 1
Advaced Science Institute, RIKEN, Hirosawa 2-1, Wako, Saitama 351-0198, Japan Institute for Solid State Physics, University of Tokyo, Kashiwa, Chiba 277-8581, Japan
2
共Presented 12 November 2008; received 17 September 2008; accepted 18 October 2008; published online 4 February 2009兲 Under electrical injection, spin accumulation occurs in lateral spin valves at the two ferromagnet/ nonmagnet interfaces, which produces a torque on the ferromagnetic electrodes, and the possibility of pure spin-current-induced magnetization reversal. Here, we generate a pure spin current in a lateral spin valve while simultaneously sweeping an external magnetic field. We observe changes to the switching properties in accordance with the effective spin torque. We also find that the spin current necessary for magnetization reversal is much lower than that required in the absence of an external field, indicative that the effective potential barrier to be overcome is lowered by the applied magnetic field. © 2009 American Institute of Physics. 关DOI: 10.1063/1.3058621兴 A pure spin current, carrying only angular momentum, can be generated electrically in lateral spin-valve 共LSV兲 structures. These are typically two ferromagnetic 共FM兲 metallic electrodes deposited along a nonmagnetic 共NM兲 nanowire, which can be a normal metal, semiconductor, or superconductor.1–4 Under electrical injection and depending on the relative magnetization of the two electrodes, spin accumulation occurs in LSVs at the two FM/NM interfaces. This can induce a torque on the FM electrodes if they are noncollinear and allows pure spin-current-induced magnetization reversal.5 Diffusion of accumulated spins from one ferromagnet, the injector, to the second detector can generate a pure spin current along the nonmagnetic wire. Spin accumulation depends on several material and growth parameters, notably the quality of the interfaces and the spin resistance. Indeed impurities at the interface will cause spin-flip scattering, which will diminish the injected spin polarization. The spin resistance is defined as RS = 2 / 共1 − P2兲S, where is the resistivity, is the spin diffusion length, P is the spin polarization, and S is the affected cross-sectional area. RS is a measure of the difficulty of spin mixing in a material; the larger the spin resistance is, the more difficult it will be for spin relaxation to occur. Spin torque switching is the action of magnetization reversal under zero applied field, and only through the propagation of carrier angular momentum from one magnetic material to another.6 If the two magnetic layers are noncollinear, then as soon as the carrier originating from FM1 reaches the differently oriented FM2, then it must align itself to FM2. As the spin angular momentum must be conserved, the transverse component of the spin is in turn absorbed by FM2, thus exerting a torque.7 Depending on the magnitude of the torque, this can lead to either magnetization reversal8 or to oscillations of the detector magnetization.9,10 However, there is a crucial difference between field-induced and currenta兲
Electronic mail:
[email protected].
0021-8979/2009/105共7兲/07D110/3/$25.00
induced magnetization reversals. In the case of the fieldinduced reversal, the applied field raises the energy of the system, so that it may transition to the other low energy state. In the case of current-induced switching, the energy gain from the spin torque and subsequently amplified precession overcomes the damping on the precession of the magnetization.11 In this study, we have applied an external field and a spin current simultaneously to a magnetic nanopillar, in a lateral spin-valve structure. All measurements were taken in the nonlocal four-probe configuration 共inset of Fig. 1兲. The sample temperature was controlled from 10 K up to RT, and a field of up to 1.5 kOe was applied in the plane of the sample, along the easy 共long兲 axis of the nanopillars. We applied an ac current 共Iac = 300 A, f = 79 Hz兲 between the Py injector and the Cu wire, and measured the voltage between the Py detector and Cu wire using a standard lock-in technique while stepping the magnetic field. We then additionally applied a dc current
FIG. 1. 共Color online兲 Field-dependence measurements of dV / dI in the nonlocal geometry for selected values of Idc, at 40 K. The curves are shifted vertically for clarity. Inset: Schematic 共top view兲 of the sample and measurement in the nonlocal four-probe geometry.
105, 07D110-1
© 2009 American Institute of Physics
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
07D110-2
Laloë et al.
Idc alongside the ac current but continuing to lock-in to Iac, thus measuring the differential resistance dV / dI. For our experiments, the injector and detector were rectangular nanopillars rather than the conventionally used micron-sized wires and pads. We were able to deposit Py nanopillars 共⬃170⫻ 80 nm2兲 on our NM Cu wire by using a mask of bilayer resist and evaporation at an oblique angle.5 This presents several advantages. First, the shape of the shadow mask enables us to deposit the Cu wire and the Py electrodes under the same continuous vacuum. This greatly limits the presence of any unwanted impurities or oxides at the interfaces between the Cu nanowire and the Py pillars. Second, the reduced lateral dimensions of the nanopillars 共as compared to nanowires兲 infer a reduced spin resistance due to the smaller cross-sectional area S.12 Finally, as we are able to deposit an injector and a detector of different thicknesses, we will be able to achieve very efficient injection while reducing the spin torque required to switch the nanopillar’s magnetization. We have already reported pure spin-current magnetization switching in such samples.5,13 We note that in order to obtain a full spin-torque hysteresis loop, it was necessary to reduce the volume of the detector. We initially perform the differential resistance measurement on a LSV in which the injector and detector were both 20 nm in thickness. The results on this sample for the case where Idc = 0 mA and a selection of dc values are shown in Fig. 1. First, we note that for Idc 艋 2.6 mA, the magnetoresistance 共MR兲 measurement was only shifted vertically by the applied dc current but the switching behavior and MR amplitude were not affected. However, for Idc = 2.7 mA and above, the upward field scan shows an anomaly, in which the occurrence of the antiparallel state is entirely suppressed. We explain this in the following way. On this section of the field-dependence scan, the magnetization of both nanopillars is initially saturated in the negative direction. Although both Py electrodes are nominally identical, we observe a large difference between their switching fields, which we attribute principally to edge roughness. For this sample and measurement, the injector has the lower switching field. As soon as the injector switches due to the external field, it is saturated along its long axis and has a negligible off-axis magnetization component. Micromagnetic simulations14 reveal that the detector, however, is in the process of magnetization switching, in the “s shape,”15 and therefore has near vertical magnetization at its edges. The spin torque, proportional to sin共⌬兲, where ⌬ is the difference in angle between the spin accumulation and detector magnetization directions, is thus immediately maximized. This provides the required energy to switch the magnetization of the detector, thus bypassing the stabilization of the antiparallel state. The decreasing field scan, however, is not affected. We repeated this measurement for the case of a negative applied dc, which inverts the polarity of the injected spin current. Here, we find that even for large values of Idc, we observe no change in the switching behavior of the LSV. This possibility is explained by an asymmetry at the injector interface, in accordance with our
J. Appl. Phys. 105, 07D110 共2009兲
FIG. 2. 共Color online兲 Switching fields of the injector 共20 nm thick; red and blue circles兲 and detector 共4 nm; black and green triangles兲, as a function of Idc, at 32 K.
previous spin torque measurements in LSVs.13,16 This asymmetry, however, allows us to exclude Oersted field effects from our analysis. We then perform the dV / dI measurements by sweeping the field while applying a dc current on a LSV for which the injector and detector are 20 nm and 4 nm thick, respectively. As previously mentioned, we have already observed a full spin-current hysteresis loop on such samples. The results are summarized in Fig. 2, where we have plotted the switching field values of the Py nanopillars as a function of Idc. In this case we observe a change of the switching field of both electrodes and for both applied current directions. We see a clear threshold of current which affects the switching behavior, although this value is not perfectly symmetrical. We first consider the lower left quadrant of the figure, where we have negative values of Idc. Both electrodes switch at lowered field values. For the case of the thick injector, through which the current is running, we attribute this change to the Oersted field in the initial stages of the switching. Since the Oersted field has a circular symmetry, it promotes the development of a vortex state and thus lowers the potential barrier of the switching.17 It is also the creation of this vortex state that causes the detector to switch at a lower magnetic field. Indeed, it provides previously inexistent perpendicular magnetization in the injector, which will amplify the excitation in the detector and aid its reversal. An increase in the temperature of the measurement from 32 to 80 K confirms this theory. In this case, we find that the dc required to alter the switching behavior of the injector is reduced by ⬃0.2 mA, as the increased thermal instability contributes to the formation of the vortex. We also observe that the detector’s switching field is not affected throughout the measured range of Idc, thus indicating as expected a decrease in the propagation of the spin accumulation due to the change in the spin diffusion length and thus the spin resistance. Finally, we compare the values of our applied current to our previously reported values for the cases where there was
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
07D110-3
J. Appl. Phys. 105, 07D110 共2009兲
Laloë et al.
no applied field.5,13 We invariably find that the electrical current required to induce switching is decreased when a magnetic field is present. This is expected, in accordance with the spin torque counteracting the damping and the magnetic field raising the energy of the system for the switching. In summary, we have measured the nonlocal differential resistance change in Py nanopillar-based LSV samples under a simultaneous magnetic field and dc current. We have found that the pure spin current generated in the Cu wire provides spin torque to the detector and stimulates its magnetization reversal at an earlier stage than in the case where there is no magnetic field. The application of a combined current and field may lead to more accessible spintronic devices, as it lowers the required spin injection efficiency. J.-B. L. acknowledges support from the Japan Society for the Promotion of Science. This work was supported in part by a Grant-in-Aid for Scientific Research in Priority Area “Creation and control of spin current” 共19048013兲 from the Ministry of Education, Culture, Sports, Science and Technology of Japan.
1
F. J. Jedema, M. S. Nijboer, A. T. Filip, and B. J. van Wees, Phys. Rev. B 67, 085319 共2003兲. 2 T. Kimura and Y. Otani, J. Phys.: Condens. Matter 19, 165216 共2007兲. 3 W. Y. Lee, S. Gardelis, B.-C. Choi, Y. B. Xu, C. G. Smith, C. H. W. Barnes, D. A. Ritchie, E. H. Linfield, and J. A. C. Bland, J. Appl. Phys. 85, 6682 共1999兲. 4 K. Miura, S. Kasai, K. Kobayashi, and T. Ono, Jpn. J. Appl. Phys., Part 1 45, 2888 共2006兲. 5 T. Yang, T. Kimura, and Y. Otani, Nat. Phys. 4, 851 共2008兲. 6 J. C. Slonczewski, J. Magn. Magn. Mater. 159, L1 共1996兲. 7 D. C. Ralph and M. D. Stiles, J. Magn. Magn. Mater. 320, 1190 共2008兲. 8 J. A. Katine, F. J. Albert, R. A. Buhrman, E. B. Myers, and D. C. Ralph, Phys. Rev. Lett. 84, 3149 共2000兲. 9 W. H. Rippard, M. R. Pufall, S. Kaka, S. E. Russek, and T. J. Silva, Phys. Rev. Lett. 92, 027201 共2004兲. 10 S. I. Kiselev, J. C. Sankey, I. N. Krivorotov, N. C. Emley, R. J. Schoelkopf, R. A. Buhrrnan, and D. C. Ralph, Nature 共London兲 425, 380 共2003兲. 11 Y. Otani and T. Kimura Elsevier Encyclopedia of Materials: Science and Technology 共Elseiver, Oxford, 2007兲. 12 T. Kimura, Y. Otani, and J. Hamrle, Phys. Rev. B 73, 132405 共2006兲. 13 T. Kimura, Y. Otani, and J. Hamrle, Phys. Rev. Lett. 96, 037201 共2006兲. 14 R. McMichael and M. Donahue, OOMMF software package, http:// math.nist.gov/oommf 15 J.-G. Zhu and Y. Zheng, in Spin Dynamics in Cofined Magnetic Structures I, edited by B. Hillebrands and K. Ounadjela 共Springer-Verlag, Berlin, 2002兲. 16 C. Heide, Phys. Rev. B 65, 054401 共2002兲. 17 G. Finocchio, I. N. Krivorotov, L. Torres, R. A. Buhrman, D. C. Ralph, and B. Azzerboni, Phys. Rev. B 76, 174408 共2007兲.
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
