gmr discovery

7/29/2019 GMR Discovery

1/4

VOLUME 61, NUIT!HER 21 PHYSICAL REVIEW LETTERS 21 NOVEMBER 1988Giant Magnetoresistance of (001)Fel(001)Cr Magnetic Snperlattices

M. N. Baibich, t') J.M. Broto, A. Fert, F. Nguyen Van Dau, and F. PetroffLaboratoire de Physique des Solides, Uni Uersite Paris-Sud, F-91405 Orsay, France

P. Eitenne, G. Creuzet, A. Friederich, and J. ChazelasLaboratoire Central de Recherches, Thomson CSF, B.P. 10, F-91401 Orsay, France

(Received 24 August 1988)

We have studied the magnetoresistance of (001)Fe/(001)Cr superlattices prepared by molecular-beam epitaxy. A huge magnetoresistance is found in superlattices with thin Cr layers: For example,with rc, =9 A, at T=4.2 K, the resistivity is lowered by almost a factor of 2 in a magnetic field of 2 T.We ascribe this giant magnetoresistance to spin-dependent transmission of the conduction electrons be-tween Fe layers through Cr layers.PACS numbers: 75.50.Rr, 72.15.Gd, 75.70.Cn

There is now considerable interest in the study of mul-tilayers composed of magnetic and nonmagnetic metalsand great advances have been obtained in the under-standing of their magnetic properties. ' Recently thetransport properties of magnetic multilayers and thinfilms have been investigated and have revealed interest-ing properties resulting from the interplay between elec-tron transport and magnetic behavior. In this Letterwe present magnetoresistance measurements on(001)Fe/(001)Cr superlattices prepared by molecular-beam epitaxy (MBE). In superlattices with thin Cr lay-ers, the magnetoresistance is very large (a reduction ofthe resistivity by a factor of about 2 is observed in somesamples). This giant magnetoresistance raises excitingquestions and moreover is promising for applications.The (001)Fe/(001)Cr bcc superlattices have beengrown by MBE on (001) GaAs substrates under the fol-lowing conditions: The residual pressure of the MBEchamber was 5&10 " Torr, the substrate temperaturewas generally around 20'C, the deposition rate wasabout 0.6 A/s for Fe and 1 A/s for Cr. This depositionrate was obtained by use of specially designed evapora-tion cells in which a crucible of molybdenum is heatedby electron bombardment. The individual layer thick-nesses range from 9 to 90 A and the total number of bi-layers is generally around 30. The growth of the super-lattices and their characterization by reflection high-energy electron diffraction, Auger-electron spectroscopy,x-ray diffraction, and scanning-transmission-electron mi-croscopy have been described elsewhere. Note that theCr (Fe) Auger line disappears during the growth of a Fe(Cr) layer. This, as well as the main features of thescanning-transmission-electron-microscopy cross sec-tions, rules out a deep intermixing of Fe and Cr. How-ever, the Auger effect, which averages the concentrationsover a depth of about 12 A, cannot probe the interfaceroughness at the atomic scale. Surface extended x-ray-absorption fine-structure experiments have been startedto probe this roughness more precisely.

The magnetic properties of the Fe/Cr superlatticeshave been investigated by magnetization and torque mea-surements. The magnetization is in the plane of thelayers and an antiferromagnetic (AF) coupling betweenthe adjacent Fe layers is found when the Cr thickness tc,is smaller than about 30 k A signature of this AF in-terlayer coupling is shown in Fig. 1: As the Cr thicknessdecreases below 30 A, the hysteresis loop is progressivelytilted. For example, with tc, =9 /[t, a field Hq=2 T isneeded to overcome the antiferromagnetic coupling andto saturate the magnetization at about the bulk Fe value.When the applied field is decreased to zero, the AF cou-pling brings the magnetization back to about zero. Ascan be seen from the variation of the low-field slopes inFig. 1, the AF coupling steeply increases when tc, de-creases from 30 to 9 /[t. The existence of such AF cou-plings has already been found in Fe/Cr sandwiches bythe light-scattering and magneto-optical measurements

FIG. 1. Hysteresis loops at 4.2 K with an applied field along[110] in the layer plane for several (001)Fe/(001)Cr superlat-tices: [(Fe 60 A)/(Cr 60 A)]5, [(Fe 30 A)/(Cr 30 A)]&0, [(Fe30 A)/(Cr 18 A)]30, [(Fe 30 A)/(Cr 12 A)]~o, [(Fe 30 A.)/(Cr9 A)]4O, where the subscripts indicate the number of bilayers ineach sample. The number beside each curve represents thethickness of the Cr layers.

2472 1988 The American Physical Society


2/4

VOLUME 61, NUMBER 21 PHYSICAL REVIEW LETTERS 21 NOVEMBER 1988

/Crg A}40

-40H,tI-20 20 40

Magnetic field {kG)FIG. 2. Magnetoresistance of a [(Fe 30 A)/(Cr 9 A)]40 su-

perlattice of 4.2 K. The current is along [110] and the field isin the layer plane along the current direction (curve a), in thelayer plane perpendicular to the current (curve b), or perpen-dicular to the layer plane (curve c). The resistivity at zerofield is 54 pA cm. There is a small diA'erence between thecurves in increasing and decreasing field (hysteresis) that wehave not represented in the figure. The superlattice is coveredby a 100-A Ag protection layer. This means that the magne-toresistance of the superlattice alone should be slightly higher.

of Grunberg et al. and by the spin-polarized low-energyelectron-diffraction experiments of Carbone and Alvara-do. ' The AF coupling between the Fe layers has beenascribed to indirect exchange interactions through the Crlayers, but a theoretical model of these interactions isstill lacking. 'The magnetoresistance of the Fe/Cr superlattices hasbeen studied by a classical ac technique on small rec-tangular samples. Examples of magnetoresistance curvesat 4.2 K are shown in Figs. 2 and 3. The resistance de-creases during the magnetization process and becomespractically constant when the magnetization is saturated.The curves a and b in Fig. 2 are obtained for appliedfields in the plane of layers in the longitudinal and trans-verse directions, respectively. The field Hp is the fieldneeded to overcome the AF couplings and to saturate themagnetization (compare with Fig. I). In contrast, fieldsapplied perpendicularly to the layers (curve c) have toovercome not only the AF coupling but also the magneticanisotropy, so that the magnetoresistance is saturated ata field higher than Hs.The most remarkable result exhibited in Figs. 2 and 3is the huge value of the magnetoresistance. For tc=9A and T-4.2 K, see Fig. 2, there is almost a factor of 2between the resistivities at zero field and in the saturatedstate, respectively (in absolute value, the resistivitychange is about 23 p 0 cm). By comparison of the re-sults for three different samples in Fig. 3, it can be seen

iiR/R (H =0)

(Fe 30 A/ Cr 18 A }3Q

0 7 (Fe 30 A/ Cr 12 A)qq

o.e(Fe 30 A/Cr 9A),o

I- /+0 I-30 I-20 I-100.5-

I10

Hs20 30 / 0Magnetic field ( k G)

FIG. 3 Magnetoresistance of three Fe/Cr superlattices at 4.2 K. The current and the applied field are along the same [110]axisin the plane of the layers.

2473


3/4

VOLUME 61, NUMBER 21 PHYSICAL REVIEW LETTERS 21 NOVEMBER 1988that the magnetoresistance is lowered when the Cr thick-ness increases. At the same time, an increase of tc,weakens the AF coupling and the saturation field Hq de-creases. Similarly, we find that both the magnetoresis-tance and Hg decrease when the temperature increases:typically, from 4.2 K to room temperature, the satura-tion magnetoresistance is lowered by about a factor of 2,awhile H~ is reduced by about 30%. We point out thatthe magnetoresistance is still very significant at roomtemperature.Summarizing the data simply, giant magnetoresis-tance effects are obtained in antiferromagnetically cou-

pled Fe/Cr superlattices by our aligning the magnetiza-tions of adjacent Fe layers with an external field. Wepropose that this magnetoresistance arises from spin-dependent transmission of the conduction electronsthrough the thin Cr layers. First, we point out that per-fect interfaces would produce only specular reflectionsand diffractions of the electron ~aves, without significantchange of the longitudinal resistivity. Significant effectson the resistivity are expected only from scattering by in-terface roughness. Second, we note that the scatteringcan be strongly spin dependent in a ferromagnetic transi-tion metal. " For Cr impurities in bulk Fe, the resistivitycross section for spin-down spin-down scattering isabout 6 times smaller than for spin-up spin-up scatter-ing (we call up and down the majority and minority spindirections, respectively). " This leads us to propose thefollowing explanation for the giant magnetoresistance ofthe Fe/Cr superlattices. Thin Cr layers (9 A= 3 latticeconstants) between thicker Fe layers having parallelmagnetizations should scatter the conduction electronsroughly as Cr impurities in bulk Fe, which implies weakinterface scattering and a high coefficient of coherenttransmission for the spin-down electrons. Thus, at H& Hs, the current is carried by the spin-down electronswith a low resistivity, p= p~&&p, where pt and p~ arethe resistivities for the spin-up and spin-down currents,respectively. At zero field, with Cr layers between twoFe layers having antiparallel magnetizations, the resis-tivity is expected to be definitely higher for two reasons.First, for the electrons of one of the Fe layers, not onlythe Cr atoms but also the Fe atoms of the antiparallellayer represent possible scattering potentials and reducethe coherent transmission. Second the spin-up and spin-down currents are averaged, which suppresses the short-circuit effect by one direction of the first case. An ap-plied field, by aligning the magnetizations, progressivelyopens the spin-down spin-down channel and lowersthe resistivity. For thicker Cr layers, or at higher tem-peratures, the probability of spin-flip scattering withinthe Cr layers increases and, by mixing the spin channels,weakens the magnetoresistance. Similar concepts havebeen introduced by Cabrera and Falicov' for the resis-tivity of Bloch walls, and more recently by Johnson andSilsbee' for the problem of spin injection from a fer-

romagnet, and their formalisms could probably be adapt-ed to be applied to the magnetoresistance of our multi-layers. This is not at all within the scope of this Letterwhich is aimed only to present experimental results andto suggest an interpretation.In conclusion, we have found a giant magnetoresis-tance in (001)Fe/(001)Cr superlattices when, for thin Crlayers (9, 12, and 18 A), there is an antiparallel couplingof the neighbor Fe layers at zero field. The highest mag-

netoresistance is observed in [(Fe 30 A)/(Cr 9 A. )14p.The resistivity is reduced by almost a factor of 2 whenthe magnetization is saturated. We interpret our resultsin terms of spin-dependent transmission between fer-romagnetic layers. The giant magnetoresistance of theFe/Cr superlattices may result from an interplay of theorientation of the Fe layers by an applied field with thespin-dependent transmission between Fe layers through aCr layer. If one considers that strongly spin-dependentconduction occurs in many ferromagnetic transition-metal alloys, " the type of magnetoresistance found inFe/Cr should be observed in other transition-metal su-perlattices. The existence of a giant magnetoresistancein Fe/Cr is promising for applications to magnetoresis-tance sensors. In the samples we have studied, the satu-ration fields are obviously too high for applications but alarge magnetoresistance at relatively small fields canprobably be obtained by thickening of the Fe layers (in agiven field, this enhances the torque on the magnetic lay-ers). Alternatively high magnetoresistance effects withweaker AF couplings should probably be observed withother couples of transition metals.This work is supported in part by the Ministere de laRecherche et de 1'Enseignement Superieur, Grants No.MRES/PMFE RE 86-50-016 and No. RE 86-50-020,and by the Direction des Recherches, Etudes et Tech-niques (Ministere de la Defense), Grant No. DRET87/1344. One of us (M.N.B.) wishes to acknowledgefinancial support from Consello Nacional de Desenvol-vimento Cientifico e Tecnologico (Brazil).

' Permanent address: Instituto de Fisica, UniversidadeFederal do Rio Grande do Sul, C.P. 15051, 91SOO PortoAlegre, Rio Grande do Sul, Brazil.'M. B. Salamon, S. Sinha, J. J. Rhyne, J. E. Cunningham,R.W. Erwin, and C. P. Flynn, Phys. Rev. Lett. 56, 259 (1986).2C. F. Majkrzak, J. M. Cable, J. Kwo, M. Hong, D. B.McWhan, Y. Yafet, and C. Vettier, Phys. Rev. Lett. 56, 2700(1986).J. R. Dutcher, B. Heinrich, J. F. Cochran, D. A.Steigerwald, and W. F. EgelhoA', J. Appl. Phys. 63, 3464(1988).4P. Griinberg, R. Schreiber, Y. Pang, M. B. Brodsky, andH. Sowers, Phys. Rev. Lett. 57, 2442 (1986); F. Saurenbach,U. Walz, L. Hinchey, P. Griinberg, and W. Zinn, J. Appl.Phys. 63, 3473 (1988).E. Velu, C. Dupas, D. Renard, J. P. Renard, and J. Seiden,

2474


4/4

VOLUME 61, NUMBER 21 PHYSICAL REVIEW LETTERS 21 NOVEMBER 1988Phys. Rev. B 37, 668 (1988).E. D. Dahlberg, K. Riggs, and G. A. Prinz, J. Appl. Phys.63, 4270 (1988).7M. Rubinstein, F. J. Rackford, V. W. Fuller, and G. A.Prinz, Phys. Rev. B 37, 8699 (1988).P. Etienne, G. Creuzet, A. Friederich, F. Nguyen Van Dau,A. Fert, and 3.Massies, Appl. Phys. Lett. 53, 162 (1988).F. Nguyen Van Dau, A. Fert, M. N. Baibich, J. M. Broto,S. Hadjhoudj, H. Hurdequint, J. P. Redoules, P. Etienne,J. Chazelas, G. Creuzet, A. Friederich, and J. Massies, in

Proceedings of the International Conference on Magnetism,

Paris, France, 1988 (to be published).' C. Carbone and S. F. Alvarado, Phys. Rev. B 36, 2433(1987)."A. Fert and I. A. Campbell, J. Phys. F 6, 849 (1976); I. A.Campbell and A. Fert, in Ferromagnetic Materials, edited byE. P. Wohlfarth (North-Holland, Amsterdam, 1982), Vol. 3, p.769.'2G. G. Cabrera and L. M. Falicov, Phys. Status Solidi (b)61, 539 (1974), and 62, 217 (1974).' M. Johnson and R. H. Silsbee, Phys. Rev. B 37, 3312, 5328(1988).

2475

gmr discovery

Documents