PHYSICAL REVIEW B VOLUME 59, NUMBER 18 1 MAY 1999-II Magnetic phases of thin Fe films grown on stepped Cr 001... Ernesto J. Escorcia-Aparicio, J. H. Wolfe, Hyuk J. Choi, W. L. Ling, R. K. Kawakami, and Z. Q. Qiu Department of Physics, University of California at Berkeley, Berkeley, California 94720 Received 8 October 1998; revised manuscript received 14 December 1998 Magnetic phases of Fe films grown on curved Cr 001 with steps parallel to 100 are studied using the surface magneto-optic Kerr effect SMOKE . We found that the atomic steps 1 induce an in-plane uniaxial magnetic anisotropy with the easy magnetization axis parallel to the step edges, and 2 generate magnetic frustration either inside the Fe film or at the Fe-Cr interface, depending on the Fe film thickness and the vicinal angle. For thickness greater than 35 Å, the Fe film forms a single magnetic domain and undergoes an in-plane magnetization switching due to the competition of the step-induced anisotropy and the Fe-Cr interfacial frustration. For thickness less than 35 Å, the Fe film forms multiple magnetic domains at low vicinal angle, and transforms into a single domain at high vicinal angle. A magnetic phase diagram in the 30­45 Å thickness range was obtained using a wedge-shaped Fe film. S0163-1829 99 02918-5 I. INTRODUCTION mechanism by an experiment in which we observe a shift of the 90° magnetization switching upon varying the step in- Since its discovery in the Co/CoO system,1 the exchange duced anisotropy. In the thin Fe film regime, we observe a bias effect was believed to occur exclusively in systems multidomain structure as a result of the magnetic frustration. where the antiferromagnet AF surface is uncompensated This multidomain phase is favored for low vicinal angle the total surface moment is nonzero . However, a recent only, and a transition into a single domain phase was ob- experiment showed that exchange bias also exists in the com- served with increasing step density. pensated Fe/FeF2(110) system2 where the total AF surface moment is zero . Using a total-energy calculation, a recent II. EXPERIMENT theoretical simulation3 was able to show the existence of the exchange bias in compensated ferromagnet F /AF systems. A Cr 001 single crystal disk of 10 mm diameter was used Moreover, this simulation predicts that the ground state of a as the substrate. Half of the crystal was kept in the 001 ferromagnetic layer on a compensated AF surface should orientation while the other half was polished into a curved have the F and AF magnetic moments aligned 90° to each shape with the step edges parallel to the 100 crystallo- other. Several groups have reported the observation of this graphic direction. The curved shape provides a continuous 90° coupling,4­7 and the physical mechanism behind this 90° range of the vicinal angle from 0° to 10°. Details of the coupling was explored by an experiment performed on the substrate preparation are presented elsewhere.7 Auger elec- Fe/stepped Cr 001 system.7 Cr is an AF with an incommen- tron spectroscopy AES was used to check the cleanness of surate spin-density wave SDW . In the 123­311 K tempera- the substrate. The AES results show that a few cycles of ture range, the SDW is transverse with the spins aligned in sputtering and annealing remove all contamination except for the same direction within each 001 plane but antialigned a small amount of nitrogen which comes from the bulk Cr. It with respect to the neighboring 001 planes.8 The 001 sur- has been reported that it is very difficult and almost impos- face of Cr, then, is uncompensated. However, when an sible to obtain an absolute nitrogen-free Cr surface.11 In the atomic step is introduced on the surface, the 001 plane literature, the low-energy electron diffraction LEED pattern directly below the surface will be exposed. Since this plane has been used as a gauge for the Cr substrate cleanness. In has spins pointing in the opposite direction as the topmost some cases, a Cr crystal with a (1 1) LEED pattern was layer, the effect of the atomic step is to compensate the AF used as an indication of a clean substrate.12 In other cases, surface at the step site. The advantage of using stepped however, the (1 1) LEED pattern was attributed to the ef- Cr 001 as the compensated AF surface is that the degree of fect of the residual nitrogen at the surface,13 and a (2 2) the magnetic frustration that occurs at the Fe/Cr interface7 LEED pattern was identified as a signature of a cleaner Cr can be made tunable by simply varying the step density or surface.14 Our Cr substrate was cleaned for 3­4 weeks with the miscut vicinal angle . The introduction of the atomic cycles of Ar ion sputtering and annealing. A (1 1) LEED steps also causes a step-induced uniaxial anisotropy9,10 that pattern was first observed after a week's worth of sputtering can be used to detect the 90° coupling. and annealing. Further cleaning of the substrate eventually In this paper, we extend our previous work7 to include resulted in a stable (2 2) LEED pattern. Thus we followed different thickness regimes to construct a magnetic phase the criteria of Ref. 14 to clean the Cr substrate. All Fe films diagram for Fe films grown on stepped Cr 001 . In Ref. 7, were grown at 480 K to provide a comparison with previous the in-plane 90° magnetization switching as a function of the work on this system, where a substrate temperature of 480 K step density was attributed to the competition between the was used during the Fe film growth to achieve a smooth film Fe-Cr 90° coupling and the step-induced magnetic anisot- surface with minimal substrate-overlayer intermixing.15 ropy. In this paper, we further confirm this competition Hysteresis loops of the Fe films were obtained with in situ 0163-1829/99/59 18 /11892 5 /$15.00 PRB 59 11 892 ©1999 The American Physical Society PRB 59 MAGNETIC PHASES OF THIN Fe FILMS GROWN ON . . . 11 893 FIG. 2. SMOKE loops for 40 Å Fe film at 140 K. The magne- tization easy axis switches from perpendicular to the step edges for 4° to parallel to the step edges for 5°. dependence is beyond the scope of this paper as our focus is the effect of the magnetic frustration. The strength of the step-induced magnetic anisotropy decreases with Fe film FIG. 1. a The splitting field (HS) of hard axis loops as a thickness (dFe). In Fig. 1 b we plot HSdFe vs for all Fe function of vicinal angle at T 480 K for several Fe film thick- films studied. All curves in Fig. 1 a roughly fall into a uni- nesses (dFe). b HSdFe vs . The universal curve indicates that HS versal curve, indicating that the step-induced anisotropy fol- follows a 1/dFe dependence. lows a 1/dFe dependence. The interfacial character of this effect shows that the step-induced anisotropy is localized to surface magneto-optic Kerr effect SMOKE measurements the step edges, and no significant strain effect contributes to using a He-Ne laser as the light source. The polarization of the anisotropy. the incident beam is within the incident plane p polarized . We then cooled the sample to 140 K to study the effect of For all films studied, no polar loops were observed so that the Fe-Cr interaction. From magnetic remanence the Fe magnetization remains in the film plane. Therefore all measurements,7 we know that there are two thickness re- hysteresis loops in this paper are from longitudinal measure- gimes for Fe on Cr 001 . For thickness above 35 Å, the ments. For measurements on the curved surface, the reflec- magnetic frustration occurs at the Fe/Cr interface and the Fe tion angle of the SMOKE laser beam was used to determine film forms a single domain. For thickness below 35 Å, the the local vicinal angle. A slit was used on the path of the magnetic frustration occurs within the Fe film and the Fe reflected beam to improve the vicinal angle resolution to film forms multiple magnetic domains. We first discuss the better than 0.25°. single-domain regime. For the purpose of completeness, we include our previous results for the hysteresis loops of a 40 Å Fe film grown on stepped Cr 001 Fig. 2 . At high vicinal III. RESULTS AND DISCUSSION angle, the Fe film forms a single domain with the magneti- zation parallel to the step edges. At low vicinal angle, the Fe The step-induced magnetic anisotropy of the Fe on curved film also forms a single domain but with the magnetization Cr 001 system has been studied by measuring the hysteresis perpendicular to the step edges. This 90° magnetization loops at 480 K which is above the Ne´el temperature (TN) of switching as a function of the vicinal angle was attributed bulk Cr (TN 311 K).7 It was found that the atomic steps to the competition between the Fe-Cr 90° coupling and the induce an in-plane uniaxial magnetic anisotropy with the step-induced magnetic anisotropy.7 At low , the Fe-Cr in- easy magnetization axis parallel to the step edges. The split- teraction dominates the step-induced anisotropy so that the ting field (HS) of the hard axis loops, which is proportional Fe magnetization is perpendicular to the step edges. At high to the strength of the step-induced magnetic anisotropy, was , the step-induced anisotropy dominates the Fe-Cr interac- found to increase with increasing . In the present work, the tion so that the Fe magnetization is parallel to the step edges. HS vs relation was studied for different Fe film thicknesses The 90° magnetization switching then occurs at a critical by growing an Fe wedge on the curved Cr Fig. 1 a . It is vicinal angle where the 90° coupling compensates the step- interesting to note that the splitting field HS is roughly linear induced anisotropy. To further confirm this competition in which is different from the Fe/stepped Ag 001 Ref. 9 mechanism, we purposely modified the strength of the step- and Fe/stepped W(001) Ref. 10 systems which show a qua- induced anisotropy and studied the corresponding change of dratic dependence of the HS on . Detailed discussion on the the critical vicinal angle. It is known that adsorp- 11 894 ERNESTO J. ESCORCIA-APARICIO et al. PRB 59 FIG. 3. SMOKE measurements at 480 K with applied magnetic FIG. 4. SMOKE measurements at 180 K for the sample in Fig. field perpendicular to the step edges for a 40 Å Fe film with right 3. The 90° magnetization switching after the Au decoration shifts to column and without left column 0.12 ML Au decoration. The a lower vicinal angle due to the enhancement of the step induced greater splitting field after the Au decoration indicates an enhance- anisotropy. ment of the step-induced magnetic anisotropy. interactions. For thicker Fe films, the Fe-Fe interaction domi- tion of submonolayer amounts of metals on stepped mag- nates the Fe-Cr interaction so that magnetic domains are netic films can change the strength of the step induced formed at the Fe-Cr interface to result in a single domain Fe anisotropy.16 We deposited 0.12 ML Au on the 40 Å Fe film film. For thinner Fe films, the Fe-Cr interaction dominates at 480 K and measured the hard-axis hysteresis loops before the Fe-Fe interaction so that magnetic domains will be and after the Au deposition Fig. 3 . The increased splitting formed inside the Fe film and result in low remanence.15 On field after the Au deposition clearly shows that the step in- the other hand, the Fe-Fe interaction energy scales as the Fe duced anisotropy has been increased. Since the Fe-Cr inter- film thickness which is independent of the vicinal angle, but action occurs at the interface, far away from the deposited the Fe-Cr interaction energy scales as the step terrace length Au, we anticipate that only the step-induced anisotropy has which is inversely proportional to the vicinal angle. Thus the been modified by the 0.12 ML Au. Therefore, as a result of the increased step-induced anisotropy, we expect a lower critical vicinal angle at which the 90° magnetization switch- ing should occur. After lowering the temperature to 180 K, we found that the critical vicinal angle for the 90° magneti- zation switching indeed shifts to a lower value than that without the Au Fig. 4 . This result further confirms that the 90° magnetization switching is a result of the competition between the Fe-Cr 90° coupling and the step-induced anisot- ropy. We then studied the thin film regime 35 Å . At low temperature these Fe films exhibit multiple magnetic do- mains on a nominally flat Cr 001 substrate.15 Figure 5 shows the hysteresis loops at 140 K for a 30 Å Fe film grown on curved Cr. Different magnetic phases can be observed at different vicinal angles. For 2.5°, the hysteresis loops exhibit splitting with low remanence for magnetic fields ap- plied both parallel and perpendicular to the step edges. Since the hysteresis loop is an averaged result within the SMOKE laser spot 0.2 mm , the low remanence in both directions indicate that the 30 Å Fe film consists of multiple magnetic domains in this vicinal angle range. This multidomain struc- ture has been identified in thin Fe films grown on nominally flat Cr 001 as a result of the presence of random atomic FIG. 5. SMOKE loops measured at 140 K for a 30 Å Fe film. steps. The appearance of multiple domains in the thin film For 2.5°, a multidomain phase is observed. For 2.5°, the regime can be understood in terms of the Fe-Fe and Fe-Cr behavior is the similar to the 40 Å film. PRB 59 MAGNETIC PHASES OF THIN Fe FILMS GROWN ON . . . 11 895 agrees with the results of the remanence experiments of Ref. 7. The behavior of C1 can be qualitatively understood in light of the picture of step induced magnetic frustration pre- sented by Berger and Hopster.15 The Fe-Cr interfacial frus- tration energy per step is proportional to the terrace length L i.e., the distance between steps , where 1/L. The energy per step for frustration within the Fe film is proportional to dFe . Therefore the multidomain structure should be favored for low dFe and low . That is why the multidomain phase appears at the lower-left corner of the phase diagram. We now look at the behavior of C2 . Figure 6 shows that C2 changes only slightly as a function of the Fe film thickness. Since the C2 transition represents the critical vicinal angle at which the frustration induced 90° Fe-Cr coupling equals the step-induced anisotropy, the weak dependence of C2 on dFe implies that the energy terms for the 90° coupling and the step-induced anisotropy must have a similar thickness depen- dence. From Fig. 1, we know that the step induced anisot- FIG. 6. Phase diagram at 140 K for Fe films grown on curved Cr ropy follows a 1/dFe dependence. Thus the 90° coupling 001 . C1 marks the transition from multidomains into single do- should also have a 1/dFe dependence. This is expected main, and C2 marks the 90° magnetization switching from perpen- since the 90° coupling comes from the Fe-Cr interfacial frus- dicular to parallel to the step edges. The arrows indicate the mag- tration and an interfacial effect should scale as 1/dFe . The netization directions of the Fe film in different regimes. twisting of the Fe moments in the normal direction of the film will modify the 1/dFe dependence7 which accounts for appearance of multiple domains inside the Fe film should be the slight variation of C2 upon increasing the Fe film thick- less favored upon increasing the vicinal angle. This behavior ness. was indeed observed for higher vicinal angle in the 30 Å Fe Finally, we would like to mention that for Fe films thinner film. For 2.5° 5°, the split loop for magnetic field ap- than 20 Å, no clear transitions similar to C1 and C2 can plied perpendicular to the step edges becomes a square loop be identified. For this reason we have not included the dFe with full remanence, indicating a transition of the Fe film 20 Å regime in the phase diagram. We speculate that in from a multidomain structure into a single-domain structure this thickness regime, the magnetic multidomains at low with the Fe magnetization perpendicular to the step edges. continuously grow in size as increases. Confirmation of For 5°, we again observe the 90° magnetization switch- this speculation relies on the domain imaging study which is ing from a direction perpendicular to one parallel to the step beyond the scope of the present work. edges as in the 40 Å Fe film. Therefore there are two transi- tions for the 30 Å Fe film upon increasing the vicinal angle : the one at lower marks the transition of the Fe film from IV. CONCLUSION multiple domains into a single domain; the second one at The experiments presented in this work serve as a confir- higher marks the 90° magnetization switching as in the 40 mation for both the frustration picture presented by Berger Å Fe film. and Hopster15 and the 90° coupling picture of Ref. 7. The To obtain a better understanding of the and dFe depen- presence of a multidomain phase at low vicinal angles for Fe dence of the two transitions, we constructed a magnetic film thickness thinner than 35 Å confirms the transfer of phase diagram using a wedged Fe film which included both the magnetic frustration from the Fe-Cr interface at higher 30 and 40 Å of Fe. For operational convenience, we define thickness to the inside of the Fe film at lower thickness. the transition vicinal angles from the remanence of the loops When the magnetic frustration is located at the Fe-Cr inter- with the magnetic field applied perpendicular to the step face, a 90° magnetization switching occurs as a result of the edges: the first transition angle C1 corresponds to a change competition between the 90° coupling and the step-induced of the remanence from zero to one, and the second transition anisotropy. By changing the strength of the step-induced an- angle C2 corresponds to a change of the remanence from isotropy, we were able to confirm this competition mecha- one to zero. We then located the C1 and C2 for different nism. We also confirmed the 1/d Fe film thicknesses (d Fe dependence of the step- Fe) to construct the phase diagram at induced anisotropy and implicitly for the 90° coupling. 140 K Fig. 6 . It should be noted that the transition between different regions is not always sharp. Any discrepancies be- This work was funded in part by the Department of En- tween the phase diagram Fig. 6 and Figs. 2 and 5 are due to ergy under Contract No. DE-AC03-76SF00098, by the Na- this fact. First we turn our attention to the C1 transition. For tional Science Foundation Grant No. DMR-9805222 and was dFe 30 Å, C1 is located at 3°. The C1 decreases rapidly also supported in part by the University of California for the with increasing dFe and goes to zero at dFe 35 Å so that for conducting of discretionary research by Los Alamos Na- dFe 35 Å the multidomain phase does not exist. This tional Laboratory. 11 896 ERNESTO J. ESCORCIA-APARICIO et al. PRB 59 1 W. P. Meiklejohn and C. P. Bean, Phys. Rev. 102, 1413 1956 . 10 Hyuk J. Choi, Z. Q. Qiu, J. Pearson, J. S. Jiang, D. Li, and S. D. 2 J. Nogue´s, D. Lederman. T. J. Moran, I. K. Schuller, and K. V. Bader, Phys. Rev. B 57, R12 713 1998 . Rao, Appl. Phys. Lett. 68, 3186 1996 . 11 F. Scheurer, P. Ohresser, B. Carrie re, J. P. Deville, R. Baudoing- 3 N. C. Koon, Phys. Rev. Lett. 78, 4865 1997 . Savois, and Y. Gauthier, Surf. Sci. 298, 107 1993 . 4 Y. Ijiri, J. A. Borchers, R. W. Erwin, S.-H. Lee, P. J. van der 12 T. G. Walker, A. W. Pang, H. Hopster, and S. F. Alvarado, Phys. Zaag and R. M. Wolf, Phys. Rev. Lett. 80, 608 1998 . Rev. Lett. 69, 1121 1992 . 5 Timothy J. Moran and Ivan K. Schuller, J. Appl. Phys. 79, 5109 13 F. Meier, D. Pescia, and T. Schriber, Phys. Rev. Lett. 48, 645 1996 . 1982 . 6 T. J. Moran, J. Nogue´s, D. Lederman, I. K. Schuller, Appl. Phys. 14 Y. Joly, Y. Gauthier, and R. Baudoing, Phys. Rev. B 40, 10 119 Lett. 72, 617 1998 . 1989 . 7 Ernesto J. Escorcia-Aparicio, Hyuk J. Choi, W. L. Ling, R. K. 15 A. Berger and H. Hopster, Phys. Rev. Lett. 73, 193 1994 . Kawakami, and Z. Q. Qiu, Phys. Rev. Lett. 81, 2144 1998 . 16 W. Weber, C. H. Back, A. Bischof, D. Pescia, and R. Allenspach, 8 E. Fawcett, Rev. Mod. Phys. 60, 209 1988 . Nature London 374, 788 1995 ; W. Weber, C. H. Back, U. 9 R. K. Kawakami, Ernesto J. Escorcia-Aparicio, and Z. Q. Qiu, Ramsperger, A. Vaterlaus, and R. Allenspach, Phys. Rev. B 52, Phys. Rev. Lett. 77, 2570 1996 . R14 400 1995 .