PHYSICAL REVIEW B, VOLUME 64, 144431 Influence of strain on the magnetocrystalline anisotropy in epitaxial CrŐCoŐPd 111... films S. Boukari, E. Beaurepaire, H. Bulou, B. Carrie re, J. P. Deville, and F. Scheurer Institut de Physique et Chimie des MateŽriaux de Strasbourg, UMR 7504, CNRS­UniversiteŽ Louis Pasteur, 23 rue du Loess, 67037 Strasbourg, France M. De Santis and R. Baudoing-Savois Laboratoire de Cristallographie, CNRS (UPR 5031) associeŽ a l'UniversiteŽ J. Fourier et a l'Institut National Polytechnique de Grenoble, 25 av. des Martyrs, 38042 Grenoble, France Received 6 February 2001; published 24 September 2001 We studied the correlations between the structure in particular strain and magnetic anisotropy in thin Co/Pd 111 films uncovered and covered with Cr. Measurements were done by grazing incidence x-ray dif- fraction and magneto-optical Kerr effect. To properly describe these correlations, one has to consider a surface magnetoelastic coefficient. We demonstrate that Cr capping leads to an enhanced anisotropy strength due to strain effects, and an extended perpendicular anisotropy thickness range due to an interface contribution. DOI: 10.1103/PhysRevB.64.144431 PACS number s : 75.70.Ak, 75.80. q, 75.30.Gw Magnetic multilayers and ultrathin films often present en- Co/Pd 111 films. In particular, the influence of the capping hanced perpendicular anisotropy that make them suitable for on the strain state in the Co layer is analyzed. application in the field of magnetic recording or storage The magnetic characterization was performed by in situ devices.1 Their anisotropy properties are usually described and ex situ magneto-optical Kerr effect MOKE . The in- phenomenologically by the NeŽel pair interaction model.2 In plane and out-of-plane strains as well as the stacking of the this picture, the tendency for the magnetization to be out-of- Co layers were determined by in situ grazing x-ray diffrac- plane results from the competition between a surface or in- tion GIXD , performed at the SUV station of the French terface anisotropy (Ks), due to the broken symmetry, and CRG-IF beamline BM32 at the European Synchrotron Ra- the volume (Kv) plus the dipolar anisotropy (Kdip). The ef- diation Facility in Grenoble.9 The chamber allows simulta- fective anisotropy is given by neous in situ film growth and diffraction measurements. The energy of the impinging beam was 11 keV and the incident angle was kept above the critical angle for total reflection. K Kv Kdip Ks /d, 1 The reciprocal lattice was described by A*, B*, and C* with A*, B* in the surface plane making an angle of 60° and where d is the thickness of the film.3 C* normal to the surface: A* B* 4 2/aPd 3 However, epitaxial films usually present a strain depen- 2.642 Ć 1 and C* 2 /aPd 3 0.9342 Ć 1. For dence with thickness so that Kv is no longer constant but more details on sample preparation and diffraction experi- depends on the thickness d because of magnetostriction.4­6 If ments, see Ref. 10. this is not taken into account, as in Eq. 1 , it results in an The Co was deposited by molecular beam epitaxy at 370 apparent contribution to Ks from the strain dependence of K on a Pd 111 single crystal to obtain smoother films than Kv , although the energy is located throughout the film and those deposited at 300 K.11 The magnetic properties of un- not only at the interfaces.7 The surface anisotropy Ks ob- covered Co/Pd 111 were measured in situ by MOKE in the tained from such an analysis must therefore be viewed as an polar geometry during the growth at 370 K. The loops show effective anisotropy. full remanence between 2.0 and 4.3 monolayers ML , indi- Strain induced modification of the anisotropy could also cating a perpendicular easy axis Fig. 1 . Upon covering the appear upon capping a film with a protective layer to per- Co layer with Cr, the perpendicular region with full rema- form ex situ magnetic measurements for instance , as it is nence is extended to 7 ML. This behavior is due to a change known that capping can have a strong effect on the under- in the effective interface anisotropy, favoring a perpendicular laying film structure.8 easy axis. However, to know whether the change is located at We see therefore that the connection between the macro- the Co-Cr interface or in the volume of the film, one has to scopic anisotropy and the structural parameters can be com- check if Cr deposition induces structural changes in the un- plex. Often one could be tempted to infer, from the depen- derlying Co film. dence of anisotropy on thickness, the structural properties at The in-plane lattice parameter of uncovered Co films the origin of this anisotropy. However such a process can grown on Pd 111 was measured in situ as a function of Co easily lead to erroneous conclusions about the structure, thickness by GIXD during the growth by locating the maxi- which in the end hinder the efforts made to engineer the mum of the so-called Co truncation rod see Fig. 2 . A rod anisotropy. denotes the intensity diffracted between Bragg points be- To disentangle the origin of the different contributions to cause of finite size effects. The position of a rod perpendicu- the anisotropy, we made a study of the interplay between the lar to the surface is related to the in-plane lattice magnetic anisotropy and the structure in Co/Pd 111 and Cr/ parameter.10,12 Since the penetration depth of the beam is 0163-1829/2001/64 14 /144431 4 /$20.00 64 144431-1 ©2001 The American Physical Society S. BOUKARI et al. PHYSICAL REVIEW B 64 144431 FIG. 3. X-ray intensity along the Co truncation rod for a 14 ML FIG. 1. Remanence divided by saturation for uncovered Co/ Cr/ 11 ML Co/Pd 111 film. The L scale is in reciprocal lattice units. Pd 111 films crosses , and covered wih Cr squares . Open squares represent the data points. The best fit represented by full circles was obtained with a 50% hcp and a 50% disordered much larger than the film thickness, an average lattice pa- stacking. Indicated is the intensity which would arise from a pure rameter of the Co film is obtained. We tried several models disordered, twin-fcc, hcp, or fcc film. to fit the relaxation with thickness d of the in-plane lattice parameter: a 1/d decrease,7 a (1/d)2/3 decrease,13 and a expansion at low Co thicknesses and strain release at higher model proposed by Basson and Ball.14 In our case, these thicknesses. To account for the Cr capping on the in-plane models do not give a satisfactory description of our data. The Co strain, we kept the same functional description for the in-plane strain is rather described by the following phenom- strain as for uncovered Co/Pd 111 and added two terms. The enological law when d di new strain is given by d 1 dc / d di , 2 d d a/d b, 3 where 9.8% is the natural misfit between Co and Pd, and with a 4.6 %* ML and b 0.5 %. We emphasize that the accounts for a residual strain. Similar descriptions of strain above formula is used as a purely phenomenological descrip- relaxation were made in the case of Ni/Cu 100 ,15 tion of strain modifications. Our point is not to analyze the Co/W 110 ,16 and Co/Cu 100 .17 The best fit is obtained with relaxation mechanism but to obtain numerical values for the 0.14, dc 0.42 ML, and di 1.72 ML. Comparable val- strain as a function of Co thickness. ues of residual strain about 1.5% were also observed on Scanning along the Co truncation rod perpendicular to the Co/Pd 111 multilayers.18 film plane makes it possible to distinguish between the dif- After Cr deposition on the Co/Pd 111 films, the in-plane ferent stacking sequences of the Co planes and to determine Co lattice parameter is modified see Fig. 2 . There is an the spacing of the planes. Using Guinier's model,19 the data are fitted with a combination of hcp, fcc and twinned fcc, and disordered contributions Fig. 3 . No significant evolution in the stacking or in the interlayer parameter deduced from the position of the maximum of the peaks is noticed upon Cr deposition. On average, the covered films are slightly ex- panded out-of-plane by about 1.2 % with respect to the bulk value for hcp Co 2.04 Ć 1 ML . According to the macro- scopic elasticity laws, an expansion both in-plane and out- of-plane is unexpected but has already been observed. It has been attributed to the defects in the film.20,21 The films are composed of about 50% hcp Co and 50% disordered Co see Fig. 3 . If the stacking sequence of an fcc structure is char- acterized by the letters ABC . . . and the one for an hcp struc- ture by ABAB . . . , then by disordered Co we mean that on top of a plane A, there is the same probability to have a plane B or C. Regarding the magnetic anisotropy, a disordered film FIG. 2. In plane strain for an uncovered open squares Co/ is equivalent to a film made half of fcc Co and half of hcp Pd 111 film and covered full squares with 14 Cr ML. The lines are fits to the data see text . The inset shows the diffracted intensity Co, so that the magnetic properties of our film will be de- around the Pd K 1 and Co K 1.6 truncation rod at H 0 and scribed by a film with 25% fcc Co and 75% hcp Co. L 0.4 for a 5.7 ML Co film. Note the change in position of the Co The modification of magnetic properties upon Cr capping truncation rod right before open squares and after full squares is therefore a complex effect. Kv is no longer constant but Cr coverage. The K scale is in reciprocal lattice units. depends on the Co thickness through the thickness depen- 144431-2 INFLUENCE OF STRAIN ON THE . . . PHYSICAL REVIEW B 64 144431 TABLE I. Tabulated material dependent parameters used to de- scribe the anisotropy. K hcp hcp hcp hcp fcc dip Kmc B1 2B3 B2 B2 106 erg/cm3 106 erg/cm3 106 erg/cm3 106 erg/cm3 106 erg/cm3 12.6 4.1a 570b 220c 770d aReference 23. bReference 24. cReference 24. dReference 25. dence of strain. To estimate the strain contribution to the effective anisotropy K, we modeled the anisotropy starting FIG. 5. Anisotropy times Co thickness versus Co thickness for a from a structural description and including parameters found Co wedge covered with Cr. The full line is a fit to the data full in the literature about magnetic properties of Co see Table squares using Eq. 4 and material dependent parameters given in I . The model result was then compared to the measured Table I. The dashed curve results from a fit with no interface mag- anisotropy, which was obtained by fitting polar hysteresis netoelastic coupling constant (Bs 0). The crossed line is obtained loops with a Stoner-Wohlfarth coherent rotation model as- with a b 0 in Eq. 3 to estimate the strain induced anisotropy suming a magnetization of 1420 emu/cm3 for the Co. The change upon Cr capping see text . loops were recorded ex situ on a wedge-shaped Co/Pd 111 film covered with Cr. To ensure coherent rotation of the mag- late the high thickness data gives an effective interface an- netization they were acquired by polar MOKE in two differ- isotropy Ks of 1.38 erg/cm2 and a volume anisotropy Kv of ent geometries: perpendicular magnetic field when the easy 4.4 106 erg/cm3. axis of magnetization was in-plane, and with a magnetic field Note that the effective volume anisotropy is close to the at 79° from the normal for a perpendicular easy axis.22 Typi- one for an unstrained hcp Co thin film see Table I , so that cal magnetization loops for both geometries are represented from the slope alone one would conclude that there is no in Fig. 4 with the best fit in full line taking into account a strain in the film. fourth order term . However, the effective anisotropy could From the detailed structural description given above, we not be determined in the thickness range from 7 to 11 ML, modeled the magnetic uniaxial anisotropy using the follow- for which there is evidence of noncoherent magnetization ing expression: rotation. On the plot K*d versus thickness, one observes a nonlin- K K hcp hcp hcp hcp dip 0.75 Kmc B1 2B3 B * 2 ear behavior at low thicknesses which is not compatible with fcc Eq. 1 Fig. 5 . This curvature shows the tendency for the 0.25*B2 /d, 4 2 Ks Bs* magnetization to go back in-plane. Using Eq. 1 to extrapo- where everything is known except two parameters which are designed in bold in the equation: Kdip is the dipolar anisot- ropy, Khcp mc the hcp Co magnetocrystalline anisotropy, Ks the NeŽel interface anisotropy, and the Bi's respectively Bs) are first order bulk respectively surface magnetoelastic cou- pling constants. The Co fcc magnetocrystalline anisotropy and the effect of roughness on the anisotropy were neglected. The convention is that a positive K denotes a perpendicular easy axis. Values used for the material dependent parameters are given in Table I. The two unknown parameters-the in- terface anisotropy Ks and the surface magnetoelastic con- stant Bs-are used to fit the data. A first attempt to fit the data was done with Bs fixed to zero. The best result dashed curve of Fig. 5 was obtained with Ks 0.08 erg/cm3. Clearly the fitting is not satisfactory, especially at low thicknesses where the curvature is opposite to the measured one. At higher thicknesses however the slope is close to the one given by the data. A second attempt was therefore needed where the two FIG. 4. Normalized magnetization curve recorded in a canted parameters Ks and Bs were free to vary during the fit. Now and polar geometry on a Co wedge covered with Cr. The uniaxial the data are well described full line of Fig. 5 with Ks anisotropy was deduced from fits to the data full lines using a 0.34 erg/cm3 and Bs 11.5 erg/cm2. Also the tendency coherent rotation model including first and second order anisotropy. for the magnetization to go back in-plane is accounted 144431-3 S. BOUKARI et al. PHYSICAL REVIEW B 64 144431 for through the negative value of Bs . The influence of the of region with full perpendicular remanence upon Cr depo- surface magnetoelastic constant is therefore opposite to the sition see Fig. 1 is therefore mainly due to an increase of interface anisotropy. Note that the as fitted interface anisot- the interface anisotropy. ropy Ks 0.34 erg/cm3 is much smaller than the one de- In summary, structural and magnetic investigation of un- duced from the simple model Eq. 1 , leading to Ks covered Co/Pd 111 films and then covered with Cr has been 1.38 erg/cm3 , which demonstrates how large the mag- performed. Through a detailed analysis of the structure of the netoelastic contribution to the effective interface anisotropy film, we were able to show up the correlations between the can be. structure and the anisotropy. Capping the Co film with Cr has To try to understand how Cr deposition changes the mag- two effects: first an increase of the interface anisotropy netic anisotropy, the function K*d is plotted with the previ- which leads to an extension of the region with full perpen- ously evaluated Ks and Bs , taking a b 0 in Eq. 3 dicular remanence, and second a deformation of the under- crossed curve of Fig. 5 to virtually hinder the deformations laying Co film which increases the strength of the anisotropy. in the magnetic film due to capping. The slope of the as To properly describe the magnetic anisotropy, in particular obtained curve is smaller than the one corresponding to the towards the low thicknesses, one has to consider a surface data. The deformation induced by Cr increases thus the magnetoelastic coefficient. strength of the anisotropy over the whole thickness range. However, the thickness where the easy axis switches from Beam time at ESRF French CRG-Interface beamline is perpendicular to parallel to the surface is nearly the same as acknowledged for the present study via the French Scientific when Cr induced deformations are included. The extension Committee. 1 M.T. Johnson, P.J.H. Bloemen, F.J.A. den Broeder, and J.J. de 13 K. Ha, M. Ciria, R.C. O'Handley, P.W. Stephens, and P. Pagula, Vries, Rep. Prog. Phys. 59, 1409 1996 . Phys. Rev. B 60, 13 780 1999 . 2 L. NeŽel, J. Phys. Radium 15, 376 1954 . 14 J.H. Basson and C.A.B. Ball, Phys. Status Solidi A 46, 707 3 See, e.g., W.J.M. de Jonge, P.J.H. Bloemen, and F.J.A. den 1978 . Broeder, in Ultrathin Magnetic Structures I, edited by B. Hein- 15 J.W. Matthews and J.L. Crawford, Thin Solid Films 5, 187 rich and J.A.C. Bland Springer-Verlag, Berlin, 1994 , p. 65; 1970 . J.-P. Renard, in Magnetism and Synchrotron Radiation, edited 16 H. Fritzsche, J. Kohlhepp, and U. Gradmann, Phys. Rev. B 51, by E. Beaurepaire, B. Carrie re, and J.-P. Kappler Editions de 15 933 1995 . Physique, Les Ulis, 1996 , p. 185. 17 W. Weber, A. Bischof, R. Allenspach, C.H. Back, J. Fassbender, 4 B. Schulz and K. Baberschke, Phys. Rev. B 50, 13 467 1994 . U. May, B. Schirmer, R.M. Jungblut, G. Gušntherodt, and B. 5 Kin Ha and R.C. O'Handley, J. Appl. Phys. 85, 5282 1999 . Hillebrands, Phys. Rev. B 54, 4075 1996 . 6 For a review of magnetoelastic effects in thin films, see D. 18 F. Hakkens, A. De Veirman, W. Coene, and F.J.A. den Broeder, J. Sander, Rep. Prog. Phys. 62, 1 1999 . Mater. Res. 8, 1019 1993 . 7 C. Chappert and P. Bruno, J. Appl. Phys. 64, 5736 1988 . 19 X-ray Diffraction in Crystals, Imperfect Crystals and Amorphous 8 H. Magnan, D. Chandesris, B. Villette, O. Heckmann, and J. Bodies, edited by A. Guinier Dover, New York, 1994 . Lecante, Phys. Rev. Lett. 67, 859 1991 ; Ch. Rath, J.E. Prieto, 20 F.J. Lamelas, H.D. He, and R. Clarke, Phys. Rev. B 43, 12 296 S. Mušller, R. Miranda, and K. Heinz, Phys. Rev. B 55, 10 791 1991 . 1997 . 21 S. Ferrer, J. Alvarez, E. Lundgren, X. Torrelles, P. Fajardo, and F. 9 R. Baudoing-Savois, M. De Santis, M.-C. Saint-Lager, P. Dolle, Boscherini, Phys. Rev. B 56, 9848 1997 . O. Geaymond, P. Taunier, P. Jeantet, G. Renaud, A. Barbier, O. 22 S.T. Purcell, M.T. Johnson, N.W.E. McGee, W.B. Zeper, and W. Robach, O. Ulrich, A. Mougin, and G. Berard, Nucl. Instrum. Howing, J. Magn. Magn. Mater. 113, 257 1992 . Methods Phys. Res. B 149, 213 1999 ; 159, 210 E 1999 . 23 R. Allenspach, J. Magn. Magn. Mater. 129, 160 1994 . 10 S. Boukari, E. Beaurepaire, H. Bulou, B. Carrie re, J.-P. Deville, F. 24 V.F. Taborov and V.F. Tarasov, Fiz. Tverd. Tela St. Petersburg Scheurer, R. Baudoing-Savois, and M. De Santis, Surf. Sci. 430, 34, 2959 1992 Sov. Phys. Solid State 34, 1588 1992 . 37 1999 . 25 H. Takahashi, S. Tsunashima, S. Iwata, and S. Uchiyama, J. 11 S. Boukari, Ph.D. thesis, UniversiteŽ Louis Pasteur, Strasbourg, Magn. Magn. Mater. 126, 282 1993 ; the value of Bfcc 2 was 1998. deduced by extrapolating the magnetostriction data measured on 12 I.K. Robinson, Phys. Rev. B 33, 3830 1986 . fcc Co-Pd alloys, and using fcc 111 13 B2 /c44 . 144431-4