PHYSICAL REVIEW B VOLUME 57, NUMBER 6 1 FEBRUARY 1998-II Structure and magnetic properties of Fe/V 110... superlattices P. Isberg Department of Physics, Uppsala University, S-751 21 Uppsala, Sweden P. Granberg Department of Materials Science, Uppsala University, S-751 21 Uppsala, Sweden E. B. Svedberg Department of Physics, Linko¨ping University, S-581 83 Linko¨ping, Sweden B. Hjo¨rvarsson and R. Wa¨ppling Department of Physics, Uppsala University, S-751 21 Uppsala, Sweden P. Nordblad Department of Materials Science, Uppsala University, S-751 21 Uppsala, Sweden Received 3 March 1997; revised manuscript received 22 October 1997 Structural, magnetic, and magnetotransport properties of Fe/V 110 superlattices have been investigated. Using Al2O3 112¯0 substrates and Mo or MoxV1 x alloy seed layers, the superlattices could be grown with a large in- and out-of-plane crystal coherence. Due to large strains, magnetoelastic effects give rise to a uniaxial in-plane magnetocrystalline anisotropy with the 001 direction as the easy axis. The anisotropy energy of the strained Fe layers was found to be of similar magnitude as the one of bulk Co. The magnetotransport properties were investigated on a series of superlattice films with the nominal structure Al2O3 /Mo (100 Å)/ Fe (23 Å)/V (4 ­ 23 Å) 20 . For V thicknesses below 15 Å, only anisotropic magnetoresistance effects are present. For larger thicknesses giant magnetoresistance effects are also present, indicating antifer- romagnetic coupling across the V interlayers. The interplay between the magnetic anisotropy, hysteresis ef- fects, and the antiferromagnetic coupling is discussed. S0163-1829 98 06506-0 INTRODUCTION range 1­14 monolayers was reported. The in-plane magnetic The magnetic properties of magnetic multilayers have anisotropy showed a clear dependence on the V thickness, arising from magnetoelastic effects associated with different been the subject of numerous studies in recent years. Mag- lattice strains. Later it was shown that Fe/V 001 superlat- netic phenomena such as oscillating exchange coupling,1 gi- tices exhibits antiferromagnetic coupling for structures with ant magnetoresistance GMR ,2 and surface anisotropy3 have thin Fe 3 ML and V thicknesses in the range of 12­14 been extensively investigated. A close relationship between ML.8 Previous investigations have shown that Fe/V 110 the structure and the magnetic properties has been demon- superlattices can be grown on single-crystal MgO 111 wa- strated, one example being the influence of the interface fers at 200 °C with a large out-of-plane crystal coherence roughness on the GMR.4 Furthermore, theoretical and ex- 400 Å Ref. 9 but no reports regarding magnetic prop- perimental studies have shown that both the magnetic ex- erties on Fe/V 110 superlattices have been published. How- change coupling across the interlayer and the magnetic inter- ever, results obtained on polycrystalline samples with 110 face anisotropy are dependent on the crystal orientation.5,6 texture, indicate a weak oscillatory antiferromagnetic The heteroepitaxial growth of superlattices with different coupling10 and a small GMR arising from hysteresis crystallographic orientations inevitably leads to lattice strain effects.11 which may vary substantially between the different growth This paper reports on the structure and magnetic proper- directions.6 Therefore, the observed orientation dependence ties of Fe/V 110 superlattices. It was found that the system of the magnetic properties could in many cases arise from a can be grown with a large in-plane and out-of-plane crystal strain induced magnetic anisotropy. To distinguish between a coherence. Magnetization measurements, using supercon- magnetoelastic effect and an intrinsic orientation dependent ducting quantum interference device SQUID magnetom- magnetic property, further investigations of the relations be- etry, show that the Fe layers exhibit an uniaxial magnetic tween the structural and the magnetic properties are required. anisotropy. We will argue that these effects arise from misfit Magnetic properties of textured Fe/V multilayers have induced strain of the constituents. Magnetoresistance mea- been investigated previously, but no orientation dependence surements also show that the system exhibits GMR for V of the intrinsic magnetic properties has been reported. In a layer thicknesses around 20 Å, indicating antiferromagnetic recent paper on single crystal bct Fe/V 001 superlattices,7 a AF coupling between successive Fe layers at those thick- ferromagnetic behavior for vanadium thicknesses in the nesses. 0163-1829/98/57 6 /3531 8 /$15.00 57 3531 © 1998 The American Physical Society 3532 P. ISBERG et al. 57 EXPERIMENT The Fe/V multilayers were fabricated in a three source ultrahigh vacuum UHV based sputtering system12 with base pressure below 1 10 9 Torr (1.33 10 7 Pa). The substrates, single-crystal Al2O3 112¯0 wafers, were ultra- sonically precleaned in isopropanol and ethanol, loaded into the deposition system and thereafter annealed at 700 °C for 20 min. High purity Ar 99.9999% gas with a partial pres- sure of 5.0 10 3 Torr was used in the sputtering process yielding typical deposition rates of 0.5 and 0.7 Å/s for V and Fe, respectively, monitored by quartz crystal microbalances. The samples were rotated 50­100 rpm during deposition to prevent thickness gradients. Some of the Fe/V multilayers were grown directly on the substrate and others on seed layers of different thicknesses and compositions. All seed layers were deposited at 700 °C. A number of samples with the same nominal structure, FIG. 1. The intensity and the FWHM in rocking curve and Al2O3 112¯0 /Mo 200 Å/Mo1 xVx alloy 200 Å/ Fe 30 Å/V 2 of the 110 Bragg peak as a function of the growth temperature 20 Å]20 , were prepared at different temperatures 20­ of the Fe 30 Å /V 20 Å multilayers grown on 330 °C in order to find the optimum growth temperature. 200 Å Mo/200 Å MoxV1 x seed layer. During the growth of the Mo1 xVx alloy seed layer x was continuously increased from 0 to 1. The purpose of using an netoresistive measurements were carried out in a Lake-Shore alloy seed layer was to gradually decrease the in-plane lattice 7225 series Susceptometer/Magnetometer system. A dc cur- parameter in order to improve the epitaxial growth of the rent of 1 mA was applied in the film plane giving a current Fe/V multilayers. density of approximately 107 A/m2 and the magnetoresis- Two samples with the nominal structures, Al2O3 tance was measured with the field directed along the in-plane (112¯0)/Mo 200 Å/ Fe 31 Å/V 17 Å 60 and Al2O3 (112¯0)/ 001 and 11¯0 directions. To investigate the influence of Mo 200 Å/Mo1 xVx alloy 200 Å/ Fe 23 Å/V 16 Å]40 , the anisotropic magnetoresistance on the magnetotransport grown at 180 °C, were used for an extensive investigation properties, measurements were performed with the current of the structural and magnetic properties. To investigate parallel and perpendicular to the magnetic field. the magnetotransport properties of Fe/V 110 , a series of samples with the following sequence, Al2O3 (112¯0)/Mo RESULT AND DISCUSSION 100 Å/ Fe 23 Å/V d Å 40 where 4 d 23 Å was pre- pared. All samples were covered with a 100 Å thick cap A. Structural properties layer of V to protect the multilayer structure from oxidation The XRD measurements revealed that the growth of Fe/V upon exposure to air. ML's directly on the Al2O3 112¯0 wafers, independent of The structural quality of the samples was investigated by the growth temperature, resulted in polycrystalline samples in situ reflective high energy electron diffraction RHEED with a fairly poor 110 texture. When using seed layers of and by conventional 2 x-ray diffraction XRD CuK Mo or Mo/Mo1 xVx , it was found that an improved crystal- radiation using a powder diffractometer with a resolution of line quality could be obtained. In Fig. 1 the results from 0.005° in 2 . The XRD measurements were carried out in a XRD measurements on Fe 30 Å /V 20 Å ML's grown at low-angle region 1°­12° in 2 as well as in a high-angle different temperatures on a Mo/Mo1 xVx seed layer are region 30°­55° in 2 around the Fe/V 110 Bragg peak. shown. As can be seen, the intensity of the 110 Bragg peak Scans to search for additional Bragg peaks were also per- is strongly dependent on the growth temperature and has a formed in a larger region 10°­120° in 2 . Reciprocal space maximum value at 180 °C. At this growth temperature, the mapping RSM was performed, using a Philips MRD sys- full width half maximum FWHM of the 110 Bragg peak tem, around the Fe/V 110 , 222 , and 310 Bragg peaks, to has a minimum value in 2 , as well as in rocking curve . determine the in- and out-of-plane lattice parameter. Correc- The temperature for optimum crystal ordering 180 °C is tions for miscut and diffractometer offsets were determined also typical for the Fe/V multilayers grown on Mo seed lay- from the 112¯0 reflection of the Al2O3 substrate and the ers. Compared to samples grown on the alloy seed layer the 110 reflection of the Fe/V multilayer. To deduce the epi- intensity of the 110 Bragg peak in these samples is slightly taxy and to determine the texture of the samples, 222 and reduced and a small increase of the FWHM in and 2 of 200 pole figures from the film as well as 112¯6 and 303¯0 the 110 Bragg peak is found. A gradual deterioration of the pole figures from the Al2O3 substrate were recorded. crystal quality with decreasing thickness of the Mo seed The magnetic properties were investigated using a Quan- layer was also observed. tum Design 5.5 T SQUID magnetometer. The samples were The structural quality of one of the samples, the Fe 31 cut in rectangular pieces with the edges of the Al2O3 sub- Å /V 17 Å multilayer grown on a 200 Å Mo seed layer at strate parallel to the 001 and 11¯0 in-plane directions of 180 °C, was thoroughly investigated using different struc- the Fe/V film. All measurements were performed with the tural characterization methods. RHEED patterns of the magnetic field applied in the plane of the samples. The mag- sample surface were measured in situ both after deposition of 57 STRUCTURE AND MAGNETIC PROPERTIES OF Fe/V . . . 3533 110 surface showed a mixture of a 2D streak pattern super- imposed on a 3D point pattern which indicates an island growth. Comparison between the lattice distance in the 001 and 11¯0 directions indicated, as in the case of Mo, a slightly larger expansion compared to Fe bulk values in the 11¯0 direction. Due to the presence of magnetic fields from the magnetrons affecting the electron beam, only compara- tive measurements of the in-plane lattice parameter were per- formed. In Fig. 2, spectra from XRD measurements on the Fe 31 Å /V 17 Å sample are plotted in the low-angle region Fig. 2 a and the high-angle region Fig. 2 b . The spectrum obtained from the low-angle XRD investigation displays sharp multilayer peaks with low intensity. From the position of the peaks, the modulation wavelength of the sample was determined to 48.1 Å, which is consistent with the result from the high-angle XRD measurements. Simulation of the reflectivity data, using the program GIXA,13 gives an average interface roughness of the order of 4 Å, which corresponds approximately to 2 atomic layers. In the high-angle region, a well defined 110 Bragg peak with relatively high intensity and well resolved satellites is found. The FWHM in 2 was determined to 0.145° which gives, using the Scherrer formula,14 an out-of-plane coherence length of 650 Å corre- sponding to 14 of the total thickness of the multilayer. No additional Bragg peaks, except for the 220 peak, were found in the region 10°­120° in 2 . Figure 3 shows reciprocal space maps of the 110 , 222 , and 310 reflections from the Fe 31 Å /V 17 Å multilayer. In the subsequent calculation of the lattice parameters, con- sideration has been given to both miscut of the substrate and misalignment of the sample in the diffractometer. The 110 peak is measured twice, once for each of the optimized asymmetric peaks 222 and 310 , without changing the tilt FIG. 2. a Low-angle reflectivity curve for Fe 31 Å /V 17 Å or rotation of the sample. From the RSM maps the atomic film grown at 180 °C on 200 Å Mo seed layer. The broader inter- plane distances for the multilayer were found to have the ference peaks arise from the Mo seed layer. Three peaks arising following values: d 110 2.056 0.003 Å, d 001 1.462 from the chemical modulation are indicated. b High-angle x-ray 0.003 Å, and d 11¯0 2.073 0.003 Å. Note that d 110 diffraction data. To the left of the Fe/V 110 Bragg peak is the Mo represents the average out-of-plane atomic plane distance of seed layer 110 Bragg peak, with corresponding Laue oscillations. Fe and V and since essentially no lattice relaxation of the The indices above the curves identify the order of the satellites. in-plane parameters of the multilayer film was observed, the d 001 and d 11¯0 value can be considered as representa- the 200 Å Mo seed layer and after the last deposited Fe layer. tive of the in-plane atomic distances of Fe and V. Thus, the The Mo 110 surface showed a high order reconstruction. same lattice distortion in the two in-plane crystal directions The 110 surface mesh of Mo was nonuniformly expanded, is found (d 11¯0 /d 001 &). This result disagrees with the with the mesh expanded more in the in-plane 001 direction result from the in situ RHEED analysis. Since an ex situ than in the 11¯0 direction. The RHEED patterns of the Fe XRD experiment reflects the average lattice parameter FIG. 3. RSM of the 110 , 222 , and 310 reflections of the Fe 31 Å /V 17 Å film. The numbers indicate the order of the satellites. The isointensity con- tours are 100, 200, 400, 1000, 2000, 4000, 6000, 10 000, 20 000, and 40 000 counts, respectively. 3534 P. ISBERG et al. 57 through the whole multilayer and the RHEED pattern is characteristic only of the surface layer s , at one particular point during the deposition, such a discrepancy is compre- hensible. To fully describe the microstructural changes throughout the thickness of the film, further investigations, using, e.g., transmission electron microscopy, are required. From the reciprocal space mapping, the in-plane crystal coherence length was determined to 300 Å. Both the in-plane and out-of-plane values 650 Å of the crystal coherence are considered as lower limits since the instrumental broadening, strain effects, and interfacial roughness all contribute to the line width. The values of the in-plane and out-of-plane crys- tal coherence are much larger than the modulation period which confirm the superlattice nature of the sample. The 200 pole figure for the Fe/V SL shows two peaks separated by 180° as expected for a 110 oriented bcc single crystal. No additional peaks were found in the region 0°­85° indicating the absence of high-angle grain boundaries. Using additional results from pole figures of the 222 Bragg peaks from Fe/V and the 112¯6 and 303¯0 Bragg peaks from the Al2O3 substrate the epitaxial relationship between film and substrate could be determined as Fe/V 110 Mo 110 Al2O3 112¯0 , Fe/V 1¯11 Mo 1¯11 Al2O3 0001 , Fe/V 11¯2 Mo 11¯2 Al2O3 11¯00 . This epitaxial relationship has also been found when grow- ing, e.g., high quality bcc Nb films onto Al FIG. 4. Reduced magnetization squares M/Ms , at T 10 K vs 2O3 112 ¯0 .15 In summary, results from the different structural charac- applied magnetic field in the in-plane 001 and 11¯0 directions for terization methods show that the superlattices have a good two Fe/V 110 superlattices. a Fe 31 Å /V 17 Å grown on Mo 200 Å and b Fe 23 Å /V 6 Å grown on Mo 100 Å . Also crystal ordering. The optimum crystal quality is obtained by plotted in a is the calculated magnetization curve solid thick line depositing the Fe/V multilayers on a Mo/Mo1 xVx seed in the 11¯0 direction. layer at a growth temperature of 180 °C. Growing the Fe/V on pure Mo seed layers yields a slightly reduced crystal qual- applied field and reaches saturation at H ity, however, giving an in- and out-of-plane crystal ordering s 330 kA/m. The Fe 23 Å /V 6 Å superlattice grown on a 100 Å Mo seed that still by far exceeds the modulation period. The Fe/V layer Fig. 4 b displays a similar uniaxial behavior, but the 110 superlattice is subject to a nonuniform strain giving a magnetization shows a curved increase with applied field. crystal structure that deviates quite significantly from cubic Such a nonlinear increase of the magnetization was found for symmetry. In view of the results from other low symmetry all samples grown on the thin seed layers. No systematic magnetic systems,16 a uniaxial magnetocrystalline anisotropy variation of the curvature or of the anisotropy field with the in the Fe/V 110 superlattice is expected. thickness of the V interlayer was found. The coercivity in the easy (H 001 c ) and the hard direction B. Magnetic properties (H 11¯0 c ) as well as the saturation field (Hs) for all samples Figure 4 shows magnetization loops measured at 10 K for in the investigation are given in Table I. There is a signifi- two Fe/V superlattices grown on different seed layers. The cant variation of the coercivities and the behavior of the two curves correspond to measurements with the magnetic magnetization loops between the different samples. The non- field applied along the two in-plane 001 and 11¯0 direc- linear increase of the magnetization can be ascribed mainly tions. As can be seen, both samples show an in-plane to structural defects. The XRD studies have shown that a thin uniaxial magnetocrystalline anisotropy with the 11¯0 direc- seed layer or a slight variation from the ideal growth condi- tion as the hard direction. For both samples, the magnetiza- tion during the film deposition introduce crystalline defects tion loop in the easy 001 direction is squarelike with a in the superlattice and results in an increased interface rough- remanent magnetization value corresponding approximately ness. These defects can introduce strain relaxation in the to the saturation magnetization. The only difference between samples leading to a distribution of anisotropy fields which the samples in the easy direction is a small variation in the results in a nonlinear increase of the magnetization with ap- coercivity. In the hard 11¯0 direction, on the other hand, a plied magnetic field. Only samples that were grown on a significant difference between the samples is seen. For the Fe sufficiently thick Mo or Mo/MoxV1 x seed layer are repre- 31 Å /V 17 Å superlattice grown on a 200 Å Mo seed sentative for the highest crystal quality and show a linear layer Fig. 4 a the magnetization increases linearly with increase of the magnetization with applied field. This result 57 STRUCTURE AND MAGNETIC PROPERTIES OF Fe/V . . . 3535 TABLE I. Magnetization data. The following parameters are plane distances in the 11¯0 and 001 directions equals the listed: the thickness of the vanadium spacer (Lv), the coercive field equilibrium bulk value d 11¯0 /d 001 &. in the easy (H 001 ) and hard (H 11¯0 ) directions, the saturation In a magnetic system that is subject to a resulting stress, c c field (Hs), and the seed layer in the respective sample. , an additional term arising from the magnetoelastic energy is added to the total anisotropy energy. In a system with H 001 H 11¯0 H cubic symmetry the total anisotropy energy E s a is given by19 c c Lv Å kA/m kA/m kA/m Seed layer E 2 2 2 2 2 2 a K 1 2 2 3 3 1 4.3 8.8 100 Å Mo 3 6.4 9.2 4.2 200 100 Å Mo 2 2 2 2 2 2 9.6 11.2 14.2 200 100 Å Mo 2 100 1 1 2 2 3 3 10.7 13 10.7 160 100 Å Mo 3 111 1 2 1 2 2 3 2 3 3 1 3 1 , 12.8 14.2 10.3 318 100 Å Mo 15 14.9 5 330 100 Å Mo 3 16 9.2 6.8 330 200 Å Mo/ where K is the first order anisotropy constant for cubic sym- 200 Å MoxV1 x metry, 001 and 111 are the magnetoelastic constants in the 17 11 16 330 200 Å Mo indicated crystalline directions, and i and i are the 17.1 26.3 9 318 100 Å Mo direction cosines for the magnetization vector and the stress, 19.3 31.4 24.7 330 100 Å Mo respectively. The Fe film is subjected to a tensile stress ( 20.3 15 14.2 330 100 Å Mo 0) both in the 11¯0 direction 1 1&, 2 1/&, 3 0 and in the 001 direction 1 2 0, 3 1 . The following energy terms, therefore, contribute to the magne- is consistent with the result from the XRD study which toelastic energy: shows that the samples grown on thick Mo or Mo/MoxV1 x seed layers give the optimal structural quality. 3 In order to quantify the anisotropic behavior, we consider E001 2 001 sin2 1 a magnetic system with a uniaxial magnetocrystalline anisot- ropy energy Ea expressed as and E 3 a K1 sin2 K2 sin4 , 1 E11¯0 4 001 111 sin2 , where K1 and K2 are the first and second order uniaxial anisotropy constants and is the angle of the magnetization where is the angle in the 110 plane between the magne- vector with respect to the easy direction of the sample. tization vector and the easy 001 direction. Using the mag- The equilibrium condition between the reduced magneti- netoelastic constants for bulk Fe, 001 20.7 10 6 and zation M/M 111 21.2 10 6,20 it can be seen that both energy terms s and an external field H applied in the hard 11¯0 direction leads to the following condition: contribute to an uniaxial anisotropy with the 001 direction as the easy axis. Assuming that the stress is the same in the 2K two energy expressions one finds that E001 E11¯0 . Identify- 1 M 4K2 M 3 ing the first order uniaxial constant K H. 2 1 in Eq. 1 with the 0M s Ms 0Ms Ms prefactor in the expression for E001 , one obtains K1 3 The M versus H curve can be fitted using the reduced mag- 2 001 . Using the value of K1 extracted from the magne- tization measurement Fig. 4 a , netization in the range 0 M/M 001 Ref. 20 and the elas- s 1, 0M s 2.2 T, and K1 tic constants21 for bulk Fe, the strain and K 001 , in the 001 di- 2 as free parameters. The fitted magnetization curve in rection can be calculated. These values give the 11¯0 direction is plotted for the Fe 31 Å /V 17 Å 001 4% to be compared with sample in Fig. 4 a . The best fit to the experimental data 001 2% determined from the diffraction experiments. A quantitative comparison between the calcu- gives K1 330 kJ/m3, K2 5 kJ/m3. The value of the anisot- lated and the experimental values of the lattice strain is ropy energy is 25 times larger than the low temperature speculative since the magnetoelastic constants value for bulk iron (13 kJ/m3) Ref. 17 and is of the same 001 and 111 , as well as the elastic constants in a thin film under consider- order of magnitude as the low temperature value of the an- able strain is expected to deviate from the bulk values. Fur- isotropy energy for cobalt (700 kJ/m3).18 thermore, Eq. 3 assumes small deviations from cubic sym- As was shown in Fig. 4, the Fe/V 110 superlattices pos- metry which is not the case in the present samples. However, sess a large in-plane uniaxial magnetocrystalline anisotropy the magnetoelastic theory qualitatively model a uniaxial be- with the 001 direction as the easy axis. This anisotropy can havior of the anisotropy with the easy axis in the 001 di- be ascribed to the reversed magnetostriction produced by the rection, as has been found from the magnetization experi- internal strain in the superlattice. Due to the lattice misfit ments. between Fe and V, the Fe layers are subjected to tensile strain both in the 001 and the 11¯0 direction. The XRD investigations have shown that the relative expansion of the C. Transport properties Fe layers in the two orientations is approximately the same The main contributions to the magnetotransport properties ( 2%) which implies that the ratio between the atomic in multilayers which include a ferromagnetic element arise 3536 P. ISBERG et al. 57 FIG. 5. Reduced electrical resistance R/R(H 0) at 10 K vs applied magnetic field for a Fe 23 Å /V 6Å superlattice. The magnetic field is applied in the film plane in the parallel H 11¯0 and I 11¯0 and in perpendicular H 001 and I 11¯0 orienta- tions. from the anisotropic magnetoresistance AMR Ref. 22 and in some magnetic multilayers also from the GMR. To quan- tify the AMR, the spontaneous anisotropic magnetoresis- tance is defined as / AMR / 0 , 4 where and are the resistivities when the magnetization direction is parallel or perpendicular to the current I. 0 is the resistivity of the thermally demagnetized state. In contrast to the AMR effect, the GMR is independent of the in-plane current direction but is, on the other hand, gov- erned by the relative orientation of the magnetization of the ferromagnetic layers in the superlattice. The magnitude of FIG. 6. Reduced electrical resistance, R/R(H 0), at 10 K vs the GMR is usually expressed by the following relation: applied magnetic field for a Fe 23 Å /V 21 Å superlattice. The magnetic field is applied in the film plane in a the parallel / GMR / , 5 (H 001 and I 001 and in b the perpendicular H 11¯0 and I 001 orientation. where is the resistivity at antiferromagnetic alignment of the magnetization in the successive ferromagnetic layers and (H 11¯0 , I 001 Fig. 6 b orientation. In both orienta- is the resistivity at saturation the magnetization in the tions the magnetoresistance decreases with increasing field, successive ferromagnetic layers are aligned parallel to the which is a signature of the GMR effect. No influence of magnetic field . For many multilayered systems / GMR AMR on the magnetoresistance could be observed. As can ( / )AMR. be seen in Fig. 6 b , the magnitude of the GMR in the per- In Fig. 5 the magnetoresistance at 10 K for the Fe 23 pendicular orientation is approximately 1.6% which is near Å /V 6 Å superlattice is plotted vs applied field in the par- the value found from the virgin sample measured in the par- allel H 11¯0 and I 11¯0 and the perpendicular allel orientation 1.9% . In the parallel orientation the mag- (H 001 and I 11¯0 orientation. In the parallel orienta- nitude of the GMR only reaches 0.7%. tion, the direction of the magnetization is continuously ro- The large difference of the GMR effect between the par- tated from perpendicular to parallel the current as the field is allel and perpendicular direction in the Fe 23 Å /V 21 Å increased from zero to the saturation field. The increase of superlattice is due to an interplay between the magnetocrys- the resistance with increasing field in this orientation is a talline anisotropy, the pinning forces and the exchange cou- signature of the AMR. When the field is applied in the easy pling that favors an antiparallel alignment of the Fe layers. In 001 direction, the magnetization is either parallel or anti- the easy direction, the magnetization curve is squarelike with parallel to the magnetic field. The relative orientation be- a coercive force of approximately 32 kA/m. The pinning tween the current and the magnetization is unchanged during forces have a strong influence on the magnetization loop and the field cycle and hence the resistivity is constant. A similar the exchange coupling across the V layers has only an influ- influence of the AMR on the total resistance, as illustrated in ence on the magnetic structure at magnetic field strengths Fig. 5, was found for all samples with V thicknesses less close to the coercive field, where a less complete antiparallel than 15 Å. alignment of the moments is achieved. When the magnetic For V thicknesses larger than 15 Å, an influence of the field is reversed after saturation in the hard direction, the GMR on the magnetoresistive properties is observed. This is magnetic moments will initially, due to the magnetocrystal- illustrated in Fig. 6 where the magnetoresistance is plotted line anisotropy, undergo a reversible rotation towards the for the Fe 23 Å /V 21 Å superlattice in the parallel easy directions. During this reversible process the magnetic (H 001 , I 001 Fig. 6 a and in the perpendicular exchange coupling may be strong enough to impose a nearly 57 STRUCTURE AND MAGNETIC PROPERTIES OF Fe/V . . . 3537 isotropy of magnetoelastic origin develops. In Fe/V 001 superlattices on the other hand, the Fe layers are symmetri- cally strained and the fourfold crystallographic and magne- tocrystalline anisotropy in-plane symmetry retains.7 In this case the anisotropy energy increases linearly with increasing strain of the Fe layers. Investigation of the magnetotransport properties on Fe/V 001 , with thick Fe,7 have shown that for V thicknesses up to 20 Å the magnetoresistance is characterized by the AMR. This finding is in contrast to the present result from the Fe/V 110 system in which the GMR effect was found for V thicknesses larger than 15 Å, indicating antiferromagnetic coupling between the Fe layers. The details of this ``orienta- tion dependence'' of the magnetic coupling in Fe/V is, how- ever, unclear. Based on the results from the structural char- FIG. 7. Magnetoresistance at 10 K vs thickness of the V inter- acterization it has been found that the interface roughness is layer DV for the Fe/V 110 superlattices. The AMR data, defined different in the Fe/V 110 and the Fe/V 001 systems, at according to Eq. 4 where 0 , are extracted from the parallel equal Fe thickness. To determine if the observed ``orienta- orientation H 11¯0 and I 11¯0 . The GMR data, defined ac- tion dependence'' of the magnetic coupling is due to the cording to Eq. 5 , are extracted from thermally demagnetized interface roughness or an intrinsic orientation effect, as sug- samples. The lines are guides to the eye. gested in a theoretical model,5 requires further investigations. complete antiparallel structure when the field is reversed to In summary, Fe/V 110 superlattices have been grown on zero. The fact that the GMR in the hard direction is of the Al2O3 112¯0 substrates, using Mo and Mo1 xVx alloy seed same magnitude as found in the virgin sample implies that layers. The lower limit of the in-plane and out-of-plane crys- the successive Fe layers form a nearly complete antiparallel tal coherence is determined to be 300 and 650 Å, respec- structure at zero magnetic field. tively, for a 3000 Å thick film. The interface roughness is In Fig. 7 the magnitude of the magnetoresistance at 10 K limited to approximately 2 monolayers. A large in-plane is plotted as a function of the thickness of the V interlayer. uniaxial anisotropy that dominates the cubic magnetocrystal- The AMR data, defined according to Eq. 4 , where line anisotropy is observed. The origin of this uniaxial an- 0 isotropy is assigned to magnetoelastic effects due to the mis- is the resistivity at zero field, are extracted from mea- surements with the applied field in the hard direction and fit strain in the superlattice. The magnetotransport parallel to the current H 11¯0 and I 11¯0 . The GMR measurements show that for V thicknesses less than 15 Å, data, defined according to Eq. 5 , are extracted from the anisotropic magnetoresistance is present. For larger thick- virgin curve for thermally demagnetised samples. For all V nesses, also giant magnetoresistance effects appear indicat- thicknesses less than 16 Å, the anisotropic magnetoresistance ing an antiferromagnetically coupled structure. The interplay governs the magnetoresistive properties. A slight decrease of between the antiferromagnetic coupling, the in-plane the AMR with increasing V thickness can be seen. For V uniaxial anisotropy and the hysteresis effects is discussed. thicknesses of approximately 16 Å also the GMR effect is observed in the magnetoresistive data. When further increas- ACKNOWLEDGMENTS ing the V thickness, the GMR dominates over the AMR. Henk van Greevenbroek is acknowledged for help in DISCUSSION AND SUMMARY evaluating the RHEED patterns and Lynnette D. Madsen for assistance with the XRD measurements. Gabriella Anderson The lattice mismatch between Fe and V causes a tuneable is also acknowledged for assistance in the sample growth. elongation of the in-plane lattice parameter of Fe in Fe/V This work has been performed within the Thin Film Consor- superlattices. For the Fe/V 110 superlattices the cubic sym- tium and financial support from NUTEK and NFR is grate- metry is broken and a large in-plane uniaxial magnetic an- fully acknowledged. 1 D. M. Edwards, J. Mathon, R. B. Muniz, and M. S. Phan, Phys. B. Dieny, T. C. Huang, and H. Lefakis, ibid. 47, 11 579 1993 ; Rev. Lett. 67, 493 1991 ; S. S. P. Parkin, N. More, and K. P. E. Fullerton, D. Kelly, J. Guimpel, I. Schuller, and Y. Bruynser- Roche, ibid. 64, 2304 1990 . aede, Phys. Rev. Lett. 68, 859 1992 . 2 B. Dieny, J. Magn. Magn. Mater. 136, 335 1994 . 5 P. Bruno and C. Chappert, Phys. Rev. Lett. 67, 1602 1991 ; 67, 3 K. Baberschke, Appl. Phys. A: Solids Surf. 62, 417 1996 ; F. J. 2592 1991 . A. Broeder, W. Hoving, and P. J. H. Bloemen, J. Magn. Magn. 6 B. D. Hermsmeier, R. F. C. Farrow, C. H. Lee, E. E. Marinero, C. Mater. 93, 562 1991 . J. Lin, R. F. Marks, and C. J. Chien, J. Appl. Phys. 69, 5646 4 G. R. Harp, S. S. P. Parkin, R. F. C. Farrow, R. F. Marks, M. F. 1991 . Toney, Q. H. Lam, T. A. Rabedeau, and R. J. Savoy, Phys. Rev. 7 P. Granberg, P. Nordblad, P. Isberg, B. Hjo¨varsson, and R. Wa¨p- B 47, 8721 1993 ; V. S. Speriosu, J. P. Nozieres, B. A. Gurney, pling, Phys. Rev. B 54, 1199 1996 . 3538 P. ISBERG et al. 57 8 P. Poulopoulos, P. Isberg, W. Platow, W. Wisny, M. Farle, B. H. Zabel, Phys. Rev. B 47, 2353 1993 . Hjo¨rvarsson, and K. Baberschke, J. Magn. Magn. Mater. 48, 480 16 A. Marty, B. Gilles, J. Eymery, A. Chamberod, and J. C. Joud, J. 1997 ; B. Hjo¨varsson, J. Dura, P. Isberg, T. Watanabe, T. J. Magn. Magn. Mater. 121, 57 1993 ; H. J. Elmers and U. Grad- Udovic, G. Andersson, and C. F. Majkratz, Phys. Rev. Lett. 79, mann, Appl. Phys. A: Solids Surf. 51, 255 1990 ; B. M. Clem- 901 1997 . ens et al., J. Magn. Magn. Mater. 121, 37 1993 . 9 J. E. Mattson, E. E. Fullerton, C. H. Sowers, and S. D. Bader, J. 17 B. Westerstrand, P. Nordblad, and L. Nordborg, Phys. Scr. 11, Vac. Sci. Technol. A 13, 276 1995 . 383 1975 . 10 S. Parkin, Phys. Rev. Lett. 67, 3598 1991 . 18 American Institute of Physics Handbook McGraw-Hill, New 11 P. Gru¨nberg, J. Barnas, F. Saurenbach, J. Fuss, A. Wolf, and M. York, 1972 , pp. 5­208. Vohl, J. Magn. Magn. Mater. 93, 58 1991 . 19 S. Chikazumi, Physics of Magnetism Wiley, New York, 1964 , 12 Peter Isberg, Ph.D. thesis, Uppsala University, 1997. p. 183. 13 P. V. de Weijer and D. K. G. de Boer, Philips J. Res. 47, 247 20 E. W. Lee, Rep. Prog. Phys. 18, 184 1955 . 1993 . 21 A. Guy and J. Hren, Elements of Physical Metallurgy, 3rd ed. 14 B. D. Cullity, Elements of X-Ray Diffraction Addison-Wesley, Addison-Wesley, New York, 1974 , p. 52. London, 1978 , p. 102. 22 I. Campbell and A. Fert, in Ferromagnetic Materials, edited by E. 15 P. Bo¨deker, A. Abromeit, K. Bro¨hl, P. Sonntag, N. Metoki, and P. Wohlfarth North-Holland, Amsterdam, 1982 , Vol. 3.