Spin filtering in ferromagnetic bilayers Henri-Jean Drouhin and Nicolas Rougemaille Citation: Journal of Applied Physics 91, 9948 (2002); doi: 10.1063/1.1476087 View online: http://dx.doi.org/10.1063/1.1476087 View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/91/12?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Magnetization reversal in ferromagnetic thin films induced by spin-orbit interaction with Slonczewski-like spin transfer torque J. Appl. Phys. 116, 133905 (2014); 10.1063/1.4897156 Three-dimensional spin structure in exchange-biased antiferromagnetic/ferromagnetic thin films Appl. Phys. Lett. 95, 092503 (2009); 10.1063/1.3216055 Electrical spin injection in multiwall carbon nanotubes with transparent ferromagnetic contacts Appl. Phys. Lett. 86, 112109 (2005); 10.1063/1.1882761 Magnetization reversal of the ferromagnetic layer in IrMn/CoFe bilayers J. Appl. Phys. 92, 6699 (2002); 10.1063/1.1518769 Magnetization dependence of training effect of exchange coupling in ferromagnet/FeMn bilayers Appl. Phys. Lett. 81, 3428 (2002); 10.1063/1.1517711 [This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP: 212.26.4.131 On: Wed, 02 Sep 2015 11:36:33 JOURNAL OF APPLIED PHYSICS VOLUME 91, NUMBER 12 15 JUNE 2002 Spin filtering in ferromagnetic bilayers Henri-Jean Drouhin and Nicolas Rougemaille Laboratoire de Physique de la Matière Condensée (UMR 7643-CNRS), Ecole Polytechnique, 91128 Palaiseau Cedex, France 共Received 6 February 2002; accepted for publication 12 March 2002兲 A general analysis of spin-polarized electron transmission through ultrathin ferromagnetic bilayers is presented. The system is analyzed in terms of density operator and a compact description of spin-filtering and spin-precession effects is given. The Sherman function, which characterizes the spin selectivity, is precisely defined. Application to spin detectors is discussed and, in particular, the two important cases where the layer magnetizations are collinear or orthogonal are studied in detail. © 2002 American Institute of Physics. 关DOI: 10.1063/1.1476087兴 I. INTRODUCTION A new field of the solid-state technology, spintronics, is presently emerging and electron devices based on the electron spin manipulation in solids may expend our capacities in key domains.1 Spin-dependent transport experiments have rapidly raised a broad interest because of the new physics involved and of their promises of application. Studies on magnetoresistance effects and related phenomena are nowadays an active field of research stimulated by industrial applications to high-density recording or magnetic sensors,2 spin-valve transistor structures have been proposed,3 highperformance magnetoresistive random access memories 共MRAMs兲 based on spin-dependent tunneling are fabricated4 and very attracting new memories, based on magnetization reversal triggered by a spin current are emerging.5 Progress in the fabrication technologies enable us to associate the properties of ferromagnetic thin films with those of metals, semiconductors, or insulators, leading to original electron devices6 and a new class of convenient spin polarimeters, making use of the solid-state technology is under development.7 This article deals with ultrathin ferromagnetic layers, described in terms of spin filters, analogous to optical polarizers. However, the electrons offer more possibilities than the photons because the electron spin polarization possibly has both longitudinal and transverse components. The inelastic electron mean free path 共IMFP兲 in metals, which characterizes the escape depth, plays a crucial role in these experiments, as in all electron spectroscopies. Its analysis in the low-energy regime 共below a few tens of eV兲 remained a puzzling question for tens of years. In Refs. 8 –10, it was shown that an accurate analytical description of the IMFP is obtained in a model based on density-of-state effects, in the range 5–50 eV above the Fermi level. This does not depend on the detail of the d bands, but only on the numbers of s and d electrons, and the various electron relaxation channels are disentangled. These results constitute the starting point of the present analysis. Hereafter, we will carefully describe spin filtering and spin precession in the density-operator formalism and we will focus on the ferromagnetic bilayer structure, which associates a ‘‘polarizer’’ and an ‘‘analyzer.’’ This spinvalve structure plays a central role in the study of new phe- nomena and presents some unusual properties, which are conceptually related to Stern and Gerlach experiments. In particular, the parameters which characterize the spin selectivity, known as the Sherman functions in spin polarimetry,11 will be carefully defined. The link between the overall Sherman function of the bilayer structure and the polarization of the transmitted beam will be clarified.12 II. SPIN PRECESSION AND SPIN FILTERS Spin-dependent transport of low-energy electrons in a ferromagnetic metal arises because majority- and minorityspin electrons have different relaxation channels due to different final densities of states in the d spin subbands. Now, consider the one-dimensional experiment where an electron beam with a longitudinal polarization 共propagating along the z axis兲 crosses a ferromagnetic layer magnetized along z 共the vectors are indicated by boldface characters兲. Referring to the initial polarization as P 0 , the polarization P of the transmitted beam is P⫽( P 0 ⫹S)/(1⫹S P 0 ). In this expression, S⫽(t ⫹⫹ ⫺t ⫹⫺ )/(t ⫹⫹ ⫹t ⫹⫺ ) is the asymmetry of the transmission coefficients. The electron transmission only depends on the relative orientation of the incident spin and the majority-spin direction in the ferromagnetic layer and t ⫹⫹ (t ⫹⫺ ) is the transmission coefficient of an up-共down-兲 spin electron when the majority spins in the ferromagnetic layer are parallel to the direction of the z quantization axis 共indicated by the first⫹index兲. Following the terminology widely used in spin polarimetry, we refer to the S parameter as the Sherman function.11 In the case where the incident beam polarization 共along the z axis, the propagation direction兲 and the layer magnetization axis 关in the (x,y) plane, say along the x axis兴 are orthogonal, the electron spin both undergoes spin filtering in the x direction, which tends to align it in the x direction, and spin precession around the x axis, in the exchange field of the ferromagnet, which results in the emergence of a y component. This effect was indeed observed by Oberli and co-workers.13 In their experiment, a longitudinally spin-polarized electron beam was injected trough a free-standing ultrathin ferromagnetic film with an in-plane magnetization. Because the primary beam is injected into the sp bands 共at a few eV above the metal Fermi 0021-8979/2002/91(12)/9948/4/$19.00 9948 © 2002 American Institute of Physics [This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP: 212.26.4.131 On: Wed, 02 Sep 2015 11:36:33 J. Appl. Phys., Vol. 91, No. 12, 15 June 2002 H.-J. Drouhin and N. Rougemaille level兲, spin precession originates from the sp-band splitting, whereas spin filtering is a consequence of the spin-dependent inelastic scattering, i.e., of the d-band splitting. An analysis of these effects was given by N. Rougemaille et al.14 A. Density-operator formalism The interest to use a density-matrix description of the spin filters was pointed out by H. C. Siegmann.15 More generally, the electron beam entering a ferromagnetic layer is described by the density operator 共1兲 ˆ the Pauli where 具S典 is the mean value of the electron spin, operator, and I the identity.16 The time evolution of the density operator in the spin filter is given by D共 t 兲 ⫽U共 t 兲 D U共 t 兲 ⫹ , 共2兲 where the evolution operator U(t) is related to the Hamiltonian H through the relation U(t)⫽exp⫺i(Ht/ប). 16 Consider a ferromagnetic layer where an orbital eigenstate with a given wave vector corresponds to two different energies E ⫹ and E ⫺ depending whether its spin state is a majority- 共⫹兲 or a minority- 共⫺兲 spin state. An electron injected at this wave vector but with a spin state which is not a pure state will undergo spin precession. As long as only precession is concerned, H is diagonal with E ⫹ and E ⫺ real eigenvalues; the energy spin splitting is ⌬E⫽E ⫺ ⫺E ⫹ ⫽ប⍀ and, as the majority spin states should correspond to a lower energy, ⍀ should be positive. Electron absorption, i.e., spin filtering, can be empirically taken into account by adding imaginary energy contributions.17 Thus, we write 冉 E ⫹ ⫺i ⫹ 0 0 E ⫺ ⫺i ⫺ 冊 共3兲 . Let us define T ⫹ (t)⫽exp⫺i(E⫹⫺i⫹)t/ប and T ⫺ (t)⫽exp ⫺i(E⫺⫺i⫺)t/ប. From Eq. 共2兲, we obtain D共 t 兲 ⫽ 冉 R⍀ 共 u兲 P⫹S , 1⫹S•P d ⫹⫹ 兩 T ⫹ 共 t 兲 兩 2 d ⫹⫺ T ⫹ 共 t 兲 T ⫺ 共 t 兲 * * T ⫹共 t 兲 *T ⫺共 t 兲 d ⫹⫺ d ⫺⫺ 兩 T ⫺ 共 t 兲 兩 2 冊 , 共4兲 where d i j (i, j⫽⫾) are the matrix elements of D. Let us denote as D⬘ the density matrix of the beam leaving the ferromagnetic layer, then the number of transmitted electrons with a majority or a minority spin is Tr(D⬘ 兩 ⫾ 典具 ⫾ 兩 ), the transmitted intensity is TrD⬘ , and the spin polarization component of the emerging beam in the ␣ direction ( ␣ ⫽x,y,z), P ␣ ⫽Tr(D⬘ ˆ ␣ )/Tr D⬘ . Note that this statistical picture is conceptually different to the coherent quantum transmission for pure spin states proposed in Ref. 13. The electrons cross the layer in the time ⫽d/ v , where d is the layer thickness and v , their group velocity. The relations 兩 T ⫹ ( ) 兩 2 ⫽exp ⫺(2⫹d/vប)⫽exp⫺(⫹ d) and 兩 T ⫺ ( ) 兩 2 ⫽exp⫺(2⫺ d/vប) ⫽exp⫺(⫺ d) connect the imaginary energy components ⫾ to the spin-dependent scattering cross sections ⫾ calculated in Ref. 9. From the D⬘ expression, it is straightforward to show that the polarization P⬘ of the emerging beam is14 共5兲 where S⫽S u, u being the unit vector parallel to the direction of the majority spins in the ferromagnetic layer and R⍀ (u) is the matrix corresponding to the composition of a clockwise spin rotation of the angle ⍀ around u 共spin precession兲 with the multiplication of the spin component in the plane normal to u by the (1⫺S2 ) 1/2 homothety ratio 共spinfilter effect兲. It is also readily checked that the transmitted intensity is I 共 u兲 ⫽I 共 1⫹S•P兲 , 1 1 ˆ, D⫽ I⫹ 具 S典 • 2 ប H⫽ P⬘ ⫽ 9949 共6兲 where I is the intensity which would be transmitted if the layer were nonmagnetic (S⫽0) or equivalently if the primary beam were unpolarized (P⫽0). B. Electron transmission through ferromagnetic bilayers Consider a ferromagnetic bilayer with an arbitrary magnetization of each layer. Since in the absence of quantum interferences the transmitted current I through the multilayer is the product of the transmitted current through each layer, I 共 u,v兲 ⫽I 共 1⫹S1 •P0 兲共 1⫹S2 •P1 兲 . 共7兲 In this expression S1 ⫽S 1 u (S2 ⫽S 2 v) is the Sherman vector of the first 共second兲 layer and Pi⫺1 (Pi ) is the polarization of the beam entering 共emerging of兲 the ith layer. I is the current which would be transmitted through the structure if the layers were not magnetized (S 1 ⫽S 2 ⫽0), which is no longer equivalent to the transmitted current for an unpolarized primary beam because the first layer acts as a polarizer. Using Eq. 共5兲, we obtain I 共 u,v兲 ⫽I 关 1⫹S 1 S 2 u•v⫹S 1 P0 •u⫹S 2 R⍀ 共 u兲 P0 •v兴 . 共8兲 Equation 共8兲 can be rewritten in the form I 共 u,v兲 ⫽I 0 共 u,v兲关 1⫹S共 u,v兲 •P0 兴 , 共9兲 where I 0 (u,v)⫽I 关 1⫹S 1 S 2 u•v兴 . Analogously to Eq. 共6兲, Eq. 共9兲 defines the overall Sherman vector S(u,v) of the structure and I 0 (u,v) is the transmitted intensity when the primary beam is unpolarized, which depends on the 兵u,v其 configuration. This dependence is closely related to a magnetoresistance effect. It can be easily shown that the last term in Eq. 共8兲 verifies the relation R⍀ 共 u兲 P0 •v⫽R⍀ 共 u兲 v•P0⬘ , 共10兲 where P0⬘ is the mirror symmetric of P0 with respect to the 共u,v兲 plane, i.e., changing the sign of the polarization component along (u⫻v) when u and v are not collinear. When u and v are collinear, the P0⬘ component in the plane perpendicular to u is arbitrary and we can take P0⬘ ⫽P0 ; this case is obvious. In the first case, the relation is derived after developing P0 on the 共u,v,u⫻v兲 basis 共which, in general, is nonorthogonal兲. The following properties have been used: [This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP: 212.26.4.131 On: Wed, 02 Sep 2015 11:36:33 9950 J. Appl. Phys., Vol. 91, No. 12, 15 June 2002 H.-J. Drouhin and N. Rougemaille R⍀ (u) u⫽u; R⍀ (u) (u⫻v)⫽ 兵 关 R⍀ (u)u兴 ⫻ 关 R⍀ (u)v兴 其 ; R⍀ (u)v•u⫽u•v. We define p(u,v) the polarization of the transmitted beam when the primary beam is unpolarized and first impinges onto the 2nd layer 共v axis兲, then crosses the first layer 共u axis兲. From Eq. 共5兲, we have p共 u,v兲共 1⫹S 1 S 2 u•v兲 ⫽S 1 u⫹S 2 R⍀ 共 u兲 v. 共11兲 Consequently, we find that Eq. 共8兲 writes 共12兲 where p⬘ (u,v) is obtained from p(u,v) by mirror symmetry, in the same way P⬘0 is obtained from P0 . When u and v are collinear, p⬘ (u,v)⫽p(u,v). This demonstrates that the overall Sherman vector S(u,v) is nothing but p⬘ (u,v). This vector is easily expressed in the set 兵 u, 关 v⫺(u•v)u兴 ,(u⫻v) 其 , which constitutes an orthogonal basis if u and v are not collinear. 关 1⫹S 1 S 2 共 u•v兲兴 S共 u,v兲 共13兲 When using a bilayer as a spin polarimeter, different combinations of the transmitted intensities, which will be easily measured, allow the determination of P0 . The following equations are of particular interest because they are directly related to the three polarization components: I 共 u,v兲 ⫹I 共 u,v̄兲 ⫽2I 关 1⫹S 1 共 P0 •u兲兴 , 共14兲 I 共 u,v兲 ⫹I 共 ū,v兲 ⫽2I 兵 1⫹S 2 共 u•v兲共 P0 •u兲 ⫹S 2 冑1⫺S 21 cos ⍀ P0 • 关 v⫺u共 u•v兲兴 其 ⫹S 2 冑 sin ⍀ P0 • 共 u⫻v兲兴 , I 共 u,v兲 ⫹I 共 ū,v兲 ⫹I 共 u,v̄兲 ⫹I 共 ū,v̄兲 ⫽4I, 共20兲 This relation allows us to determine (P0 •u) once S(u,v) is known. It is easy to verify that I 0 共 u,v兲 2 ⫺ 冉 ⌬I 共 u,v兲 共 P0 •u兲 冊 2 ⫽I 0 共 u,v̄兲 2 ⫺ 冉 ⌬I 共 u,v̄兲 共 P0 •u兲 冊 2 ⫽I 2 共 1⫺S 21 兲共 1⫺S 22 兲 共22兲 冉 共23兲 so that one obtains 冊 ⌬I 共 u,v̄兲 2 I 0 共 u,v̄兲 2 2 . 2 ⫽S共 u,v 兲 1⫺ I 0 共 u,v兲 ⌬I 共 u,v兲 2 This relation allows us to determine the two Sherman functions S(u,u) and S(u,ū) for parallel and antiparallel layer magnetizations, only from intensity measurements, with no need of independent characterization of the bilayer. In that sense, the structure appears as ‘‘self-calibrated.’’ D. The bilayer with perpendicular magnetizations This case, where u•v⫽0, has been discussed in Ref. 14. There, Eqs. 共14兲–共17兲 yield I 共 u,v兲 ⫹I 共 u,v̄兲 ⫽2I 关 1⫹S 1 共 P0 •u兲兴 , 共24兲 共15兲 I 共 u,v兲 ⫹I 共 ū,v兲 ⫽2I 关 1⫹S 2 冑1⫺S 21 cos ⍀ 共 P0 •v兲兴 , 共16兲 I 共 u,v兲 ⫹I 共 ū,v̄兲 ⫽2I 关 1⫹S 2 冑1⫺S 21 sin ⍀ P0 • 共 u⫻v兲兴 , 共26兲 共17兲 I 共 u,v兲 ⫹I 共 ū,v兲 ⫹I 共 u,v̄兲 ⫹I 共 ū,v̄兲 ⫽4I. I 共 u,v兲 ⫹I 共 ū,v̄兲 ⫽2I 关 1⫹S 1 S 2 共 u•v兲 1⫺S 21 I 共 u,v兲 ⫺I 共 ū,v̄兲 ⫽I 共 P0 •u兲关 S 1 ⫹S 2 共 u•v兲兴 2 and consequently 1⫺ ⫽ 共 S 1 ⫹S 2 u•v兲 u⫹S 2 冑1⫺S I2 cos ⍀ 关 v⫺ 共 v•u兲 u兴 ⫹S 2 冑1⫺S 21 sin ⍀ 共 u⫻v兲 . ⌬I 共 u,v兲 ⫽ 共19兲 S 1 ⫹S 2 共 u•v兲 ⌬I 共 u,v兲 ⫽ 共 P •u兲 ⫽S共 u,v兲共 P0 •u兲 . 共21兲 I 0 共 u,v兲 1⫹S 1 S 2 共 u•v兲 0 I 共 u,v兲 ⫽I 0 共 u,v兲关 1⫹p共 u,v兲 •P0⬘ 兴 ⫽I 0 共 u,v兲关 1⫹p⬘ 共 u,v兲 •P0 兴 , I 共 u,v兲 ⫹I 共 ū,v兲 ⫽I 关 1⫹S 1 S 2 共 u•v兲兴 ⫽I 0 共 u,v兲 , 2 where ū⫽⫺u(v̄⫽⫺v). Note that only two layers may allow to determine the three components of the incident polarization. Other combinations, obtained from intensity differences are also useful, such as the following: I 共 u,v兲 ⫺I 共 ū,v̄兲 ⫽2I 兵 关 S 1 ⫹S 2 共 u•v兲兴共 P0 •u兲 ⫹S 2 冑1⫺S 21 cos ⍀ P0 • 关 v⫺u共 u•v兲兴 其 . 共18兲 C. The bilayer with collinear magnetizations In this case, we have u⫽⫾v. The bilayer in the collinear geometry allows measuring the polarization component along the magnetization axis, P0 •u. This case has been analyzed in detail in Ref. 9. The simplest way to perform a polarization measurement makes use of Eqs. 共16兲 and 共18兲, which write 共25兲 共27兲 The structure allows to determine the three components of the polarization, as if there were a third ‘‘virtual’’ layer magnetized along the (u⫻v) axis, provided the Sherman function is known. This additional calibration can be performed either by starting with a primary beam of given polarization or by measuring the polarization of the transmitted beam when an unpolarized primary beam impinges on the flipped structure 关see Eq. 共13兲兴. It has been shown in Ref. 14 that, with a proper choice of materials and thickness, the spin sensitivity of such a structure can be as high as 0.6 in the three directions, thus competing with the best existing polarimeters. III. CONCLUSION The efficiency of ferromagnetic thin films as spin filters was experimentally demonstrated in Ref. 7. There, a low- [This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP: 212.26.4.131 On: Wed, 02 Sep 2015 11:36:33 J. Appl. Phys., Vol. 91, No. 12, 15 June 2002 energy 共a few eV兲 spin-polarized electron beam emitted from a GaAs photocathode under optical pumping18 was impinging onto a free-standing Au/Co/Au multilayer under ultrahigh vacuum. The transmitted current was energy analyzed and Sherman functions as high as 0.6 were measured. The case of ferromagnetic bilayers with collinear magnetizations is reported in Refs. 19 and 20, in complete agreement with the above analysis. Going a step further towards application requires deposing the films on convenient substrates. The possibility to grow spin filters on semiconductors was demonstrated in Ref. 21, in the case of Fe on n-type GaAs. There, a spin-polarized electron beam was injected into a Pd/Fe/ GaAs structure and the current collected in the semiconductor was detected. The system appears analogous to a transistor where the emitter is spatially separated from the base/ collector part. A large spin asymmetry in the collected current, corresponding to a Sherman function of about 0.6, was measured when the ‘‘base’’ and the ‘‘collector’’ were maintained at the same potential. The absolute current asymmetry ⌬I, which corresponds to the difference of the transmitted current for two opposite polarizations of the primary beam obviously do not depend on the ‘‘base’’ access resistance. On the contrary, it was observed that the relative asymmetry ⌬I/I 0 , where I 0 is the averaged collector current, decreases when the access resistance increases. This is not surprising, because the electrons which are not transmitted through the ferromagnetic layer accumulate and polarize the structure, resulting in a current flow with no memory of the primary beam polarization, and emphasizing the importance of the boundaries conditions, which lead from ballistic filtering to the spin diffusion regime where the transport is governed by a chemical-potential diffusion equation.22,23 The electrons captured by the ferromagnetic layer transfer their charge, but also their spin momentum, which may lead to magnetization switching,5 although this may also result from the torque exerted by the ballistically transmitted spin current on the magnetization.24,25 The complete microscopic description of the bilayer structure with noncollinear magnetizations in the diffusion regime may bring valuable information, and these effects are presently under investigation. Finally, let us emphasize that we have seen evidence of H.-J. Drouhin and N. Rougemaille 9951 the potential of ferromagnetic bilayers as three-dimensional spin polarimeters and that a convenient and physical description has been given. ACKNOWLEDGMENT H.J.D. thanks the Délégation Générale pour l’Armement for support. J. M. Kikkawa and D. D. Awshalom, Nature 共London兲 397, 139 共1999兲. E. Vélu, C. Dupas, D. Renard, J.-P. Renard, and J. Seiden, Phys. Rev. B 37, 668 共1988兲. 3 M. Johnson, J. Magn. Magn. Mater. 140–144, 21 共1995兲. 4 J. Daughton, J. Appl. Phys. 81, 3758 共1997兲. 5 J.-E. Wegrowe, D. Kelly, T. Truong, Ph. Guittienne, and J.-Ph. Ansermet, Europhys. Lett. 56, 748 共2001兲. 6 S. Datta and B. Das, Appl. Phys. Lett. 56, 665 共1990兲. 7 H.-J. Drouhin, A. J. van der Sluijs, Y. Lassailly, and G. Lampel, J. Appl. Phys. 79, 4734 共1996兲. 8 H.-J. Drouhin, Phys. Rev. B 56, 14 886 共1997兲. 9 H.-J. Drouhin, Phys. Rev. B 62, 556 共2000兲. 10 H.-J. Drouhin, J. Appl. Phys. 89, 6805 共2001兲. 11 J. Kessler, Polarized Electrons 共Springer, Berlin, 1985兲. 12 For a review, see H.-J. Drouhin, in Handbook of Infrared, edited by H. Henini and M. Razeghi 共to be published兲. 13 D. Oberli, R. Burgermeister, S. Riesen, W. Weber, and H. C. Siegmann, Phys. Rev. Lett. 81, 4228 共1998兲. 14 N. Rougemaille, H.-J. Drouhin, G. Lampel, Y. Lassailly, J. Peretti, and A. Schuhl, J. Appl. Phys. 共to be published兲. 15 H. C. Siegmann, 5th European Conference on Atomic and Molecular Physics, Edinburgh, 1995, in Selected Topics on Electronic Physics, edited by D. M. Campbell and H. Kleinpoppen 共Plenum, New York, 1996兲, p. 221. 16 R. Balian, Du Microscopique au Macroscopique 共Ellipses, Paris, 1982兲; pp. 85 and 69. 17 C. Cohen-Tannoudji, B. Diu, and F. Laloë, Mécanique Quantique 共Hermann, Paris, 1977兲, p. 939. 18 H.-J. Drouhin, C. Hermann, and G. Lampel, Phys. Rev. B 31, 3872 共1985兲. 19 H.-J. Drouhin, C. Cacho, G. Lampel, Y. Lassailly, J. Peretti, and A. J. van der Sluijs, Proceedings of Low energy polarized electron workshop, edited by Y. A. Mamaev, S. A. Starovoitov, T. V. Vorobyeva, and A. N. Ambrazhei 共PES Lab, St. Petersburg, 1998兲, p. 79. 20 C. Cacho, Y. Lassailly, H.-J. Drouhin, G. Lampel, and J. Peretti, Phys. Rev. Lett. 88, 066601 共2002兲. 21 A. Filipe, H.-J. Drouhin, G. Lampel, Y. Lassailly, J. Nagle, J. Peretti, V. I. Safarov, and A. Schuhl, Phys. Rev. Lett. 80, 2425 共1998兲. 22 T. Valet and A. Fert, Phys. Rev. B 48, 7099 共1993兲. 23 J. E. Wegrowe, Phys. Rev. B 62, 1067 共2000兲. 24 L. Berger, J. Appl. Phys. 89, 5521 共2001兲. 25 W. Weber, S. Riesen, and H. C. Siegmann, Science 291, 1015 共2001兲. 1 2 [This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP: 212.26.4.131 On: Wed, 02 Sep 2015 11:36:33