Journal of Magnetism and Magnetic Materials 192 (1999) 258-262 Magnetoresistance, magnetization and FMR study of Fe/Ag/Co multilayer film C. Birlikseven , C. Topacli *, H.Z. Durusoy , L.R. Tagirov , A.R. Koymen , B. Aktas Faculty of Engineering, Department of Physics, Hacettepe University, Beytepe 06532, Ankara, Turkey University of Texas at Arlington, Arlington, TX 76019, USA Gebze Institute for Advanced Technologies, 41410 Cayirova-Gebze/Kocaeli, Turkey Received 10 June 1998 Abstract The polycrystalline [Fe/Ag/Co/Ag];3 asymmetric multilayer film was prepared by the UHV magnetron sputtering method on silicon. In-plane magnetization measurements showed structured hysteresis loops. Magnetoresistance (MR) measurements revealed giant magnetoresistance effect with magnitudes in 0.14-0.21% range at room temperature. The saturation magnetizations and the interaction between layers were studied by ferromagnetic resonance and revealed an indistinguishably weak interlayer coupling from out-of-plane geometry of measurements. The MR data are interpreted based on incomplete domain alignment model for polycrystalline magnetic films. 1999 Elsevier Science B.V. All rights reserved. PACS: 75.60; 75.70 Keywords: FMR; GMR; Magnetic films; Magnetization 1. Introduction the conduction electron mediated exchange, never- theless show an appreciable change of resistance up There has been a great interest to investigate the to few percents, which are commonly called giant magnetoresistance of multilayers composed of Fe-, magnetoresistance (GMR). The importance of the Ni- and Co-based ferromagnets alternated with above systems is that they exhibit the GMR effect noble metals like Cu or Ag [1-6]. These multi- at small values of magnetic field (typically 5-50 G), layers, being the so-called uncoupled systems in the which make them promising sensors for magnetic sense that the magnetic layers are not coupled by read-write head applications. In this paper we re- port the magnetoresistance, magnetization and ferromagnetic resonance (FMR) studies of * Corresponding author. Tel.: #90 4 235 25 51; fax: #90 4 235 25 50. the polycrystalline [Fe(20 As)/Ag(40 As)/Co(20 As)/  On leave from Kazan State University, Kazan 420008, Ag(40 As)];3 multilayer films at room temper- Tatarstan, Russian Federation. ature. 0304-8853/99/$ - see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 4 - 8 8 5 3 ( 9 8 ) 0 0 3 0 2 - 3 C. Birlikseven et al. / Journal of Magnetism and Magnetic Materials 192 (1999) 258-262 259 2. Results 2.1. Sample preparation The films were deposited by DC magnetron sput- tering method on single crystalline Si(1 0 0) sub- strate in a UHV system at room temperature with growing rate of about 0.15 As/s Ag layers of 20 As thickness were used as buffer and top protective layers. Iron and cobalt layers were grown of 20 As thickness, the Ag spacer layer was 40 As thick to ensure the absence of conduction electron mediated exchange interaction between magnetic layers. Structural investigation made on the scanning elec- tron microscope JEOL JEM-1200EX revealed the polycrystalline growth of the films with grain size &300-400 As. 2.2. Magnetization Fig. 1. VSM measured magnetization (panel (a)) and magne- Magnetization measurements by a sensitive vi- toresistance MR(H) measurements (panel (b)) for [Fe(20 As)/ brating sample magnetometer (VSM) LakeShore Ag(40 As)/Co(20 As)/Ag(40 As)];3 multilayer films at room tem- 7300 were made with in-plane geometry for the perature. Solid curve on the panel (b) displays the MR results for magnetic field. They revealed the structured hyster- unannealed sample, the dotted curve shows MR(H) after anneal- esis loop shown in Fig. 1a. The rounded shape of ing in Ar(95%)H(5%) atmosphere at 300°C during 10 min. the loop indicates the incomplete magnetic mo- ment alignment within the layers because of mo- saicity of in-plane easy-axes of crystallites, that is nique with current and potential leads being atta- expected for our polycrystalline films. The coer- ched by the springed contacts via the silver paint civity field of the main step is H K$26 G, which dots. The resistivity R(H) of the samples in the we referred to the iron layers, is markedly smaller magnetic field H was measured in standard in- than the coercivity field for the 50 As thick Fe layer plane geometry with current and magnetic field in Ref. [5]. There are also three visible shoulders on lying in the film plane being perpendicular to each higher fields, the central one of them has the co- other. The GMR MR(H) (percents) as a function of ercivity field H +90 G, which is slightly smaller the field has been defined as than that seen in the hysteresis loop for 50 As thick Co layer presented in Ref. [5]. We attribute those R(H)!R(H) features of the hysteresis loop to the responses of MR(H)" ;100, (1) R(H the Co layers of our triply repeated Fe/Ag/Co/Ag ) structure. Overall view of the magnetization loop is where H qualitatively similar to Ref. [5] and may be ex- K500 G is the saturation field. The re- sults of the measurements are displayed on the plained by the weighted superposition of hysteresis panel (b) of Fig. 1. They clearly show hysteretic loops of each of the materials. behavior of MR with the field for maximum deriva- tive of MR(H) correlating closely with H 2.3. Magnetoresistance K $26 G obtained from the magnetization measure- ments (panel (a) of Fig. 1). The magnitude of The stripe-shaped samples with dimensions the MR(H) results presented is about 0.21% for 100 m;3 mm were prepared by lithography tech- the unannealed sample. We have made the heat 260 C. Birlikseven et al. / Journal of Magnetism and Magnetic Materials 192 (1999) 258-262 treatment with our samples, according to Hylton et al. [3], but annealing did not lead to increase of the MR magnitude (see the dashed curve on Fig. 1b and figure caption). For our long-stripe samples we expect also the contribution of anisotropic mag- netoresistance (AMR) to MR(H), that is why we have measured magnetoresistance with current and magnetic field being parallel to each other, but lying in the film plane. This resulted in approxim- ately 30% reduction of MR(H) with respect to the previous case when current and magnetic field were perpendicular. This reduction may be considered as the estimate of AMR effect in our samples. 2.4. FMR Ferromagnetic resonance experiments were car- ried out on BRUKER-EMX X-band ESR spec- trometer at "9.79 GHz at room temperature. The samples were mounted on a goniometer which allowed them to rotate with 1.0° step in out-of- Fig. 2. FMR fields for resonance (panel (a)) and layers magnetiz- plane geometry with DC magnetic field changing ations and canting angles (panel (b)) versus angle of the DC magnetic field measured from the film normal. Panel (a) symbols its angle & from the film normal ( &"0) to the show the experimental results, curves display the results of the film plane ( &"90°) and AC field lying always in theory. Panel (b) dashed and dotted curves show the equilibrium the film plane. The measurements showed overlap- angles of magnetic moments with respect to the film normal, ped resonance lines in the main domain of mea- solid line gives the canting angle between magnetizations of Fe sured angles, and two separate lines at angles close and Co layers. to the film normal. These two lines were attributed to the FMR signals coming from the Fe and Co layers constituting our multilayer system. We de- Here K is uniaxial anisotropy normal to the film veloped the computer procedures for the decompo- plane, M is the true saturation magnetization at sition of the spectra to two lines, the results for the current temperature, E   term may include any angular dependence of the resonance fields for the type of magneto-crystalline anisotropy, it was set to individual lines are shown on the panel (a) of Fig. be equal to zero for our polycrystalline samples. 2 by solid square and opened circle symbols (see the The general ferromagnetic resonance condition [7] figure caption). The linewidths for the resonance together with the condition for equilibrium, given lines were: H by the zeros of the first angular derivatives of !K587 G and H$ K303 G at E with respect to the angle , determine the FMR &"0, and H K101 G for the overlapped line at resonance field H &"90°.   as a function of the angle &. The FMR results are analyzed using a coordi- The effective magnetization 4 M can be obtained nate system in which is the polar angle measuring from the fitting of the calculated curve to the ex- the deviation of magnetization vector M from the periment. The results of the analysis are given on film normal. The free energy density is given by the upper panel of Fig. 2 by dashed and dotted lines, they show good agreement with experimental E"!(M ) H)#(2 M!K) cos #E  , (2) data. We obtained the values of the effective mag- which defines the effective magnetization netizations of the layers: (4 M )$ "16.43 kG and (4 M 4 M )!"13.94 kG. The values ( / )$ K "4 M!2K/M. (3) 3.3294 kG and ( / )!K3.0767 kG have been C. Birlikseven et al. / Journal of Magnetism and Magnetic Materials 192 (1999) 258-262 261 obtained upon fitting ( "2 , is the gyromag- by Heinrich and Cochran [11]) have discussed in netic ratio), which correspond to spectroscopic g- detail possible contributions of magnetostriction factors, g$ K2101 and g!K2.2734 for Fe and Co and NeŽel [12] mechanism to the uniaxial surface layers, respectively. anisotropy. First one comes from the strain due to The calculations give the possibility to determine the lattice mismatch between the magnetic layer the equilibrium magnetization angle for each of and the spacer (or substrate) and was found to be the layers and canting angle between the magneti- insufficient to explain the observed values of per- zations of Co and Fe layers as a function of the DC pendicular anisotropy, as the bulk magnetoelastic magnetic field angle &. The results for the magneti- coefficient has been used for the quantitative es- zation angles are displayed on the panel (b) of timation. The NeŽel mechanism attributes the sur- Fig. 2. face uniaxial anisotropy to arise from the abrupt breaking of local symmetry at the interface, and was found to give correct sign and order of magni- 3. Discussion tude, if the enhancement by the interface disloca- tions is taken into account. Another mechanism of Note first of all, that we have performed calcu- 4 M reduction is the roughness of interface [13]. lations as in Fig. 2 with the assumption that the The magnetostatic energy associated with rough- magnetic layers are noninteracting. From the lower ness always leads to reduction of effective 4 M, it panel we can learn that in the field angle range can be important for our samples, because the 3-8°, the angle between magnetizations (canting common roughness of the sputtered films of about angle), which is mainly due to different demagne- two monolayers is comparable with total film tization fields, reaches its maximum value *13°. If thickness &20 As (6-7 monolayers). It is worth to an appreciable magnetostatic interaction exists be- note that Eq. (5) of Ref. [13] is not appropriate for tween canted magnetizations, then, according to quantitative estimations in the case of ultrathin our estimations, it should manifest itself by the films, because it is derived for the roughness much appearance of resonances as the maximum canting smaller than the film thickness. According to Eq. (3) angle is reached. The experimental data on angular the energy of effective uniaxial anisotropy being dependence of the resonance field do not reveal the interpreted as the surface one K influence of interlayer interaction. Thus, we may "K/d$ , can be estimated as (K conclude that the strength of the magnetostatic )$ K0.89 erg cm\ at d$ "20 As, which is close to the values 0.69 and 0.81 erg cm\ field is considerably lower than the FMR field for for the Fe/Ag interfaces quoted in Table 1 of Ref. resonance. [11]. The value of g-factor g The value for 4 M $ K2.101 is only of the iron layers differs slightly greater than the corresponding one for the markedly from the bulk value K21.6 kG at room bulk iron g temperature [8], quoted in Ref. [9], but according $ K2.09 [8]. The value of (4 M to their measurements on the single-crystal iron )!"13.94 kG for the co- balt layers also differs well from the bulk value films on GaAs substrate with variable Fe layer (4 M thickness d )!"17.8 kG [8]. In the investigation of Co $ , upon changing d$ from 120 As down films grown on MgO and Al to 16 As, (4 M O substrates [14] )$ decreased from K17.5 to a substantial reduction of the saturation moment K7.5 kG. Prinz et al. [9] attributed the experi- upon decreasing the film thickness in the range mentally observed reduction of the effective mag- 350-120 As was found. According to Ref. [11] the netization to the influence of perpendicular surface uniaxial perpendicular anisotropy for Co layers in anisotropy, arising at the interface with substrate. Co/Au and Co/Cu superlattices contains substan- In a recent investigation of the sputtered iron films tial constant (bulk) contribution and a surface term on MgO substrate [10] a decrease of the effective proportional to 1/d 4 M from near the bulk value at the thickness !. As our samples have the con- stant thickness of Co layers, we restrict our estima- d$ "500 As to (4 M )$ K16.0 kG at dK"30 As tion only by quoting the value of (K was found. Goryunov et al. [10] (see also the review )!K2.7;10 erg cm\. The value of g-factor g!K2.2734 lies 262 C. Birlikseven et al. / Journal of Magnetism and Magnetic Materials 192 (1999) 258-262 well beyond the quoted value gK2.18 for bulk hcp board. The research of L.R.T. at Hacettepe Univer- Co [8]. In a recent investigation of Co/Cr multi- sity, Ankara is supported by TU BITAK within layers [15] with Co layers thicknesses t! around NATO-CPC AFP. C.T. and H.Z.D. acknowledge 20 As the monotonous increase of g-factor value TU BITAK for supporting the research through from g!K2.20 at t!"34 As to g!K2.26 at TBAG-1271 and NATO-B2. t!"12 As have been observed, which correlates with our observation of increased g-factor value for the thin Co film (t!K20 As) with respect to the References bulk value. The magnetoresistance data can be qualitatively [1] B. Dieny, V.S. Speriosu, S.S.P. Parkin, B.A. Gurney, explained in the multidomain model with the exis- D.R. Withoit, D. Mauri, Phys. Rev. B 43 (1991) tence of certain fraction of mutually oppositely 1297. [2] W.R. Pratt Jr., S.-F. Lee, J.M. Slaughter, R. Lolee, oriented, counterpart domains in different layers P.A. Schroeder, J. Bass, Phys. Rev. Lett. 66 (1991) [5]. There exist two configurations for the current 3060. flow across the layers: between two domains of [3] T.L. Hylton, K.R. Coffey, M.A. Parker, J.K. Howard, opposite orientation and two domains of aligned Science 261 (1993) 1021. orientation of magnetizations. If the domains are [4] D. Miyauchi, S. Araki, Appl. Phys. Lett. 63 (1993) 1702. larger compared to mean spin scattering length and [5] Y.I. Idzerda, C.-T. Chen, S.F. Cheng, W. Vavra, G.A. the spacer thickness, these two configurations will Prinz, G. Meigs, H.-J. Lin, G.H. Ho, J. Appl. Phys. 76 have essentially different resistivities. Then the (1994) 6525. magnetoresistance rises and drops will be confined [6] C. Cowache, B. Dieny, A. Chamberod, D. Benizri, F. to the coercivity fields at which the magnetizations Berthet, S. Auffret, L. Giacomoni, S. Nossov, Phys. Rev. B 53 (1996) 15027. of the layers experience reversal. Fig. 1 clearly con- [7] H. Suhl, Phys. Rev. 55 (1955) 555. firms that physical picture: the maxima of the MR [8] Numerical Data and Functional Relationships in Science curve for the unannealed sample correspond to the and Technology, Landolt-Bošrnstein, New Series, vol. magnetization reversal paths for the Fe layer mag- III/19a, Springer, Heidelberg, 1986. netization. Moreover on the MR(H) curves, there [9] G.A. Prinz, G.T. Rado, J.J. Krebs, J. Appl. Phys. 53 (1982) 2087. clearly are the shoulder-like features, which we at- [10] Yu.V. Goryunov, N.N. Garifyanov, G.G. Khaliullin, I.A. tribute to the reversal fields for the Co layers. We Garifullin, L.R. Tagirov, F. Schreiber, Th. Mušhge, H. believe that the multidomain model adequately re- Zabel, Phys. Rev. B 52 (1995) 13450. flects the polycrystalline nature of our samples. [11] B. Heinrich, J.F. Cochran, Adv. Phys. 42 (1993) 523. [12] L. NeŽel, J. Phys. Radiat. 15 (1954) 376. [13] C. Chappert, P. Bruno, J. Appl. Phys. 64 (1988) 5736. Acknowledgements [14] Yu.V. Goryunov, G.G. Khaliullin, I.A. Garifullin, L.R. Tagirov, F. Schreiber, P. Bošdeker, Ch. Morawe, Th. Muš hge, H. Zabel, J. Appl. Phys. 76 (1994) 6096. This work was supported by the Welch Founda- [15] F. Schreiber, Z. Frait, Th. Zeidler, N. Metoki, W. Donner, tion and Texas Higher Education Coordinating H. Zabel, J. Pelzl, Phys. Rev. B 51 (1995) 2920.