JOURNAL OF APPLIED PHYSICS VOLUME 87, NUMBER 9 1 MAY 2000 Temperature dependent magnetic domain structure in ultrathin Fe films on Cr 100... H. Hopstera) Department of Physics and Astronomy, University of California, Irvine, California 92697 Magnetic microscopy is used to study the temperature dependent magnetization structures in 2 nm Fe films on Cr 100 . Above the Cr NeŽel temperature, the Fe films can be magnetized into a single domain state. When the films are cooled below the NeŽel temperature the Fe magnetization has a tendency to turn perpendicular in-plane resulting in a spatially varying magnetization direction. The resulting magnetization structures are highly reproducible. The tendency of the Fe magnetization to rotate is attributed to frustration due to atomic steps. It is suggested that the local angle of magnetization rotation reflects the average step density. © 2000 American Institute of Physics. S0021-8979 00 65608-6 Exchange coupling at ferro/antiferromagnetic FM/AF 2 nm were grown on bulk Cr 100 see Ref. 3 for growth interfaces can lead to interesting magnetic properties. The conditions . In our SEMPA we can measure the two in-plane effect of exchange biasing has been known for a long time.1 magnetization components in a medium-energy Mott detec- It has become of increasing technological importance re- tor. See Fig. 1 for a schematic of the measurement geometry. cently. However, on an atomic scale the effect is not well A coil allows us to apply magnetic fields in the x direction. understood and interest in the subject has recently been The SEMPA images are acquired in zero magnetic field. The increasing.2 The combination of well characterized samples spatial resolution is approximately 500 nm. The sample can with novel magnetic probes promises to shed new light on be cooled by LN2 or heated indirectly on the SEMPA stage. this old problem. This study reports on the temperature de- Above the NeŽel temperature of Cr (TN 311 K) the Fe pendent magnetization in 2 nm Fe films on Cr 100 . The films can be magnetized into a single domain state. Figure 2 Fe/Cr systems have been studied extensively in conjunction shows the evolution of the magnetization starting with the with exchange coupling in Fe/Cr/Fe layered structures. An magnetized state in the top panel. The left column shows the advantage of this system is that it offers very high quality x component and the right column the y component. When epitaxial samples. Although, to our knowledge, exchange bi- the film is cooled the single-domain state x magnetization asing has not been reported for Cr as the antiferromagnet a breaks up into irregular magnetization structures and a strong recent magneto-optical Kerr effect MOKE study revealed y component develops with spatial variations on the scale of unusual temperature dependent properties of the hysteresis micrometers to tens of micrometers. A polarization analysis curves of thin Fe films on Cr 100 .3 In particular, the coer- shows that the magnetization magnitude is conserved and cive field and the remanent magnetization in films thinner only an in-plane rotation takes place see Fig. 3 of Ref. 4 for than 5 nm showed anomalies that were clearly associated typical angular distributions . When the films are warmed to with the magnetic ordering of Cr. Thus, this is an example in above TN the magnetization direction relaxes into the easy- which the FM properties are influenced or even dominated axis directions of Fe x and y axes . This is clearly visible as by the AF magnetic structure due to strong interface cou- a sharpening and increased contrast of the images. When, pling. In the present study we use magnetic microscopy to however, this structure is cooled down again the image re- examine the temperature dependent magnetization structures to show how a uniform in-plane magnetization breaks up into a domain structure when cooled below the AF Cr or- dering temperature. This article gives additional information to what has already been reported in a recent letter on the subject.4 The experiments were performed in a new ultrahigh vacuum system that allows for molecular beam epitaxy sample growth and surface characterization low-energy electron diffraction and Auger and in situ magnetic micros- copy by secondary electron microscopy with polarization analysis SEMPA . A scanning tunneling microscope STM has just been added to the system which will allow us to characterize atomic scale structure and study correlations be- tween morphology and magnetic structure. Ultrathin Fe films FIG. 1. Geometry of the experiment. Two in-plane magnetization compo- nents x and y are measured. The films can be magnetized by a magnetic a Electronic mail: hhopster@uci.edu field pulse in the x direction. 0021-8979/2000/87(9)/5475/3/$17.00 5475 © 2000 American Institute of Physics Downloaded 20 Mar 2001 to 148.6.169.65. Redistribution subject to AIP copyright, see http://ojps.aip.org/japo/japcpyrts.html 5476 J. Appl. Phys., Vol. 87, No. 9, 1 May 2000 Hebert Hopster FIG. 3. Images (50 m 50 m) of the two in-plane magnetization com- ponents as in Fig. 2 at low temperatures. The second row was taken after the sample had been warmed up above TN , then magnetized into a single domain, and then cooled again. allel to each other. We suggest that in the present case we have an effective perpendicular coupling already at the single AF/FM interface. We have recently added an STM to our system which allows us to characterize the surface topography of our FIG. 2. Effect of cooling and warming on the magnetic structure of a 2 nm Fe film. The images are of an area about 50 m 50 m. Left panels show samples. Fig. 4 shows one of the first STM images obtained the x component and the right panels the y component. The series start from for our sample. It shows a 500 nm 500 nm randomly chosen the top with a fully magnetized sample above TN . The second row was area on a well prepared clean Cr surface before deposition of taken at low temperature while the third row is after warming to above TN . an Fe film. As expected, atomic terraces in the range of tens of nanometers are observed, a range typical for high-quality single-crystal metal substrates. One has to keep in mind that verts to the previous low-temperature ``fuzzy'' image in the magnetic structures observed are much larger microme- which the Fe magnetization directions varies locally. Thus, ters than these terrace widths. Thus, the magnetic structures we have three different magnetization structures: 1 The ``average'' over many terraces and it is suggested that they fully magnetized single domain state after magnetizing are a reflection of the average of the local atomic step den- above TN ; 2 The low temperature state with a locally, con- sity. tinuously varying magnetization direction; 3 The state after warming in which the magnetization lies only along the easy magnetization axes of the Fe film. Even though the magnetization structures are very ir- regular it is interesting to note that there is a high degree of reproducibility. This is shown in Fig. 3. When the magnetic structure is erased by warming and magnetizing and then the film is cooled down again we find a high degree of repro- ducibility of the magnetic structure. What causes certain areas to rotate their magnetization direction upon cooling while other areas remain close to their original direction? It is well known that roughness can lead to frustrated magnetization structures at AF/FM interfaces. The Fe­Cr exchange interaction cannot be minimized in the presence of atomic steps without inducing magnetic transi- tion regions either in the Fe or the Cr see Ref. 3 . This can lead to 90° alignment of the magnetic moments. This has been quite well documented in exchange coupling between FM films through interlayers, where this phenomenon is called ``biquadratic coupling'' as opposed to ``bilinear cou- FIG. 4. STM image of a 500 nm 500 nm area of the Cr 100 substrate after pling,'' in which case the moments align parallel or antipar- the cleaning and annealing procedures. The area was randomly chosen. Downloaded 20 Mar 2001 to 148.6.169.65. Redistribution subject to AIP copyright, see http://ojps.aip.org/japo/japcpyrts.html J. Appl. Phys., Vol. 87, No. 9, 1 May 2000 Hebert Hopster 5477 magnetized. As the temperature is lowered further it cost more energy to accomodate the mismatch in the Cr because of the increased Cr magnetic order. If the step density is low enough the mismatch stays in the Cr. However, if the step density is large it costs too much energy to accomodate the mismatch in the Cr. Also, Fe domains on the scale of the atomic terraces are energetically unfavorable and therefore the Fe magnetization tends to turn 90°. The resulting mag- netization structure is then a compromise between the fluc- tuating interface exchange interaction and the stiffness of the Fe magnetization. For the real sample, of course high and low step densities vary continuously and this leads to the spatially varying magnetization direction observed. We want to stress that the Fe magnetization turns away from the direction of the Cr surface moments. The direction of the Cr moments was, of course, determined by the Fe magnetization when the sample was cooled through TN . That means that the Fe magnetization should always turn away from its original direction at high temperature, inde- FIG. 5. Schematic of the temperature dependent magnetic structure of the Fe films and the Fe/Cr interface: Left column: low step density area; right pendent of the actual initial magnetization direction. This is column: high step density area. Top panels: Above TN ; fully magnetized Fe observed experimentally from which one can conclude that it film and no bulk magnetic order in Cr; center panels: Magnetic order devel- is really an exchange coupling to the Cr surface moment ops in Cr induced by the Fe magnetization. The magnetic mismatch is ac- alignments that drives the Fe reorientation and not, e.g., commodated in the Cr. Lower panels: At low step density locations left the Fe magnetization remains fixed and the mismatch is still accommodated in anisotropies induced by step orientations. the Cr. At high step densities right the Fe starts turning in-plane towards In summary, we have shown that thin Fe films on 90°. Cr 100 develop a nonuniform magnetic structure as the tem- perature is lowered sufficiently below TN . It is suggested that the driving mechanism is the atomic scale roughness that Figure 5 shows, schematically, the proposed magnetiza- leads to a locally varying in-plane turning of the magnetiza- tion structures as the temperature is lowered. The left column tion because of varying steps densities and frustration. In the shows an area with a small step density and the right column future we hope to be able to correlate magnetization images is at high step density. What exactly ``high'' and ``low'' with atomic terrace structure studied by STM. mean will have to be determined in a systematic survey of The STM was funded by the Army Research Office step densities by STM in the future. Above TN the Fe films through an Instrumentation Grant. are magnetized and there is no long-range magnetic order in the Cr, except for possibly some interface magnetization in- duced by the Fe. As the sample is cooled below TN the Cr magnetization develops, induced by the Fe due to the Fe­Cr 1 W. H. Meiklejohn and C. P. Bean, Phys. Rev. 102, 1413 1956 . exchange coupling, i.e., the Cr moments order, starting at the 2 For a recent review on exchange bias see: J. NogueŽs and I. K. Schuller, J. Fe interface. In this situation the magnetization mismatch is Magn. Magn. Mater. 192, 203 1999 . 3 A. Berger and H. Hopster, Phys. Rev. Lett. 73, 193 1994 . accommodated in the Cr and the Fe remains essentially fully 4 H. Hopster, Phys. Rev. Lett. 83, 1227 1999 . Downloaded 20 Mar 2001 to 148.6.169.65. Redistribution subject to AIP copyright, see http://ojps.aip.org/japo/japcpyrts.html