Journal of Magnetism and Magnetic Materials 198}199 (1999) 489}492 Magnetization reversal in out-of-plane magnetized Ni/Cu(1 0 0) "lms A. Bauer*, E. Mentz, G. Kaindl Institut fu(r Experimentalphysik, Freie Universita(t Berlin, D-14195 Berlin, Germany Abstract Field-induced magnetization reversal in Ni/Cu(1 0 0) "lms with perpendicular anisotropy is studied in situ by magneto-optical Kerr e!ect (MOKE) and Kerr-microscopy. We discuss the time dependence of the reversal process and determine the Barkhausen volume. Surface defects are found to generate metastable 3603 domain walls that strongly in#uence the magnetization reversal process. 1999 Elsevier Science B.V. All rights reserved. Keywords: Magnetization reversal; Magnetic domains; Thin magnetic "lms 1. Introduction magnetic anisotropy. In contrast to most thin-"lm sys- tems, Ni/Cu(1 0 0) shows a spin-reorientation transition Recently, much attention has been paid to magnetiz- from in-plane (below +7 ML) to out-of-plane magnetiz- ation-reversal processes in ultrathin magnetic "lms in- ation ('7 ML) [3]. From dynamical magnetization duced by external magnetic "elds [1,2]. For the magnetic measurements, we can determine < "eld parallel to the easy axis of magnetization, the rever- and the activation energy for a Barkhausen step, E sal process generally takes place by domain nucleation . Furthermore, macro- scopic surface defects are found to pin the domain walls: and subsequent domain growth by domain-wall motion. The walls wind around these defects and form metastable Both processes are thermally activated and depend on 3603 domain walls behind them. the strength of the external magnetic "eld. Variations of the domain-wall energy due to surface defects and "lm morphology (locally varying "lm thickness) represent 2. Experimental details barriers for domain-wall motion and may slow down the magnetization reversal process by several orders of mag- The experiments were performed in a UHV system nitude. For the same reason, domain-wall motion occurs equipped with low-energy electron di!raction (LEED), in discontinuous steps, and the average volume that is scanning tunneling microscopy (STM), MOKE, and magnetically reversed in one step is the so-called Bar- Kerr-microscopy [4,5]. The ex situ electro-polished khausen volumea, < . Cu(1 0 0) single-crystal surface was cleaned in situ by In the present work, we have studied the dynamics of several sputter-anneal cycles [4,5]. The Ni "lms were magnetization reversal in out-of-plane magnetized 15- deposited at room temperature by e-beam evaporation of monolayers (ML) thick Ni/Cu(1 0 0) "lms. Ni/Cu(1 0 0) is a Ni wire, and the "lm thickness was measured with known for its anomalous thickness dependence of the a quartz balance. In situ polar MOKE and Kerr-microscopy were car- ried out at room temperature. Domain images were re- * Corresponding author. Tel.: #49-30-838-6162; fax: #49- corded with a CCD camera by "rst taking a background 30-838-6560; e-mail: bauer@physik.fu-berlin.de. image of the magnetically saturated "lm that was 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 1 2 0 2 - 5 490 A. Bauer et al. / Journal of Magnetism and Magnetic Materials 198}199 (1999) 489}492 Fig. 1. (a) Polar MOKE hysteresis loop for 15 ML Ni/Cu(1 0 0) with coercive "eld of +12.5 Oe (":H!). (b) Relative magnetization M/M1 as a function of time for various external magnetic "elds H. Prior to magnetization reversal, the "lm was magnetized with H"40 H!. (c) Same curves as in (b) but scaled with t/t(H). The linear dependence of ln(t) on H is shown in the inset. Lower left inset: (400 nm) STM image of a clean Cu(1 0 0) surface. (d) Magnetization curves for H"!0.6 H! for two di!erent initial magnetization "elds H. Curve B is "tted with an exponential function. subsequently subtracted in real time from the images surface or interface steps represent obstacles for the taken during magnetization reversal. The lateral resolu- domain-wall motion, which is also found for other tion of the Kerr-microscope is +3 m. thin-"lm systems [7]. A series of domain images, taken in the process of magnetization reversal at room temperature at the same 3. Results and discussion 15-ML-thick Ni/Cu(1 0 0) "lm, is shown in Fig. 2(a) (images 1}5). Basically, one domain front with reversed In Fig. 1(a), the rectangular-shaped polar MOKE hys- magnetization (dark area) moves across the "eld of view. teresis loop of a 15-ML-thick Ni/Cu(100) "lm is shown. It By changing the magnetic "eld, the speed of the domain was recorded with a "eld sweep rate of about 10 Oe/s, wall can be adjusted to any desired value. Since the easy resulting in a coercive "eld of +12.5 Oe (":H!). For axes of magnetization within the Ni-"lm plane are the higher (lower) sweep rates, the coercive "eld increases +1 1 0, directions [3], the Bloch-type domain walls are (decreases) since the magnetization-switching time de- preferably oriented along these directions. At several pends exponentially on the external magnetic "eld H. In points on the surface the domain walls are pinned, wind Fig. 1(b), the time dependence of the magnetization rever- around the defects, and "nally merge behind them. If the sal is plotted for several "eld values. The shapes of the magnetic "eld is again reversed after magnetic saturation plotted curves are nearly identical, which is seen by was obtained (image 5), domain nucleation takes place plotting the curves with di!erent time scales t/t(H) (see along a line network that corresponds to the lines along Fig. 1(c)); here, t is de"ned by M(t)"0. As is seen in the which the domain walls merged in the "rst reversal pro- inset of Fig. 1(c), t depends exponentially on the external cess. Even though it cannot directly be resolved with the magnetic "eld as is expected for thermally activated Kerr microscope, the present data imply that behind the domain-wall motion: tJexp[(E ! M1H< )/k ¹]. defects the merging domain walls do not annihilate but M1 is the saturation magnetization within the domains. form 3603 walls. The reason for the formation of meta- With ¹"300 K and M1"0.45 T [6], < is deter- stable 3603 domain walls is that the sense of rotation mined from the slope of ln t(H) as < "1.9;10 nm. of the magnetization within the two merging Bloch- It corresponds to a Barkhausen area (< divided by the like 1803-domain walls is identical, which is sketched "lm thickness) of 0.073 m, which is of the same order as in Fig. 2(b). the average size of a terrace on the Cu(1 0 0) surface It is interesting to note that the speed of domain-wall (see STM image in the inset of Fig. 1(c)). Obviously, motion is strongly enhanced if magnetization reversal A. Bauer et al. / Journal of Magnetism and Magnetic Materials 198}199 (1999) 489}492 491 Fig. 2. (a) Magnetic-domain images taken in the process of magnetization reversal on a 15-ML Ni/Cu(1 0 0) "lm (images 1}5). After an apparent one-domain state was obtained in a magnetic "eld close to H! (image 5), the external magnetic "eld was reversed (images 6}9). Images 6}9 were recorded in magnetic "elds signi"cantly smaller than H!. (b) Schematic drawing of the formation of a 3603 wall. The arrows in the domain wall indicate the sense of rotation of the magnetization vector within the Bloch-like wall. starts from a 3603 domain wall network (images 6}9 in wall network. While the magnetization is almost un- Fig. 2(a)). It needs up to a factor 2 higher magnetic "elds changed within the "rst few seconds in curve A, it to observe the same propagation speed if no 3603 domain switches within (1 s in curve B. This e!ect can be walls are present (images 1}5). This Kerr-microscopy explained in the following way: Formation of 3603 do- result is also re#ected by dynamical MOKE measure- main-walls costs energy and slows down the magnetiz- ments. In Fig. 1(d), two magnetization curves are shown ation-reversal process; the e!ective activation energy that were recorded for the same external magnetic "eld E is enhanced. On the other hand, if the 3603 domain H"!0.6H!. Prior to recording of curve A, the "lm walls have already formed, the magnetization-reversal was magnetized in a magnetic "eld H"40 H! that was process is basically the time-reversed process of the 3603 su$cient to destroy the metastable 3603 domain walls. domain-wall formation: The walls open up with an area of For curve B, on the other hand, the "lm was magnetized reversed magnetization in between two 1803 walls. In this in H"1.4 H!, which obviously did not a!ect the 3603- case, E is reduced since pinning by defects does not occur. 492 A. Bauer et al. / Journal of Magnetism and Magnetic Materials 198}199 (1999) 489}492 A detailed description of the magnetization curves in References Fig. 1(b) and (d) will be published elsewhere [8]: As a result, it turns out that the exponential time depend- ence observed in the presence of 3603 domain walls (curve [1] D. Sander et al., Phys. Rev. Lett. 77 (1996) 2566. B in Fig. 1(d)) can be described with a model based on [2] A. Moschel et al., Phys. Rev. Lett. 77 (1996) 3653. [3] K. Baberschke, M. Farle, J. Appl. Phys. 81 (1997) Fatuzzo's theory [9]. With this model, the intrinsic ac- 5038. tivation barrier height for a Barkhausen step can be [4] E. Mentz et al., Mat. Res. Soc. Symp. Proc. 475 (1997) determined as E 0.7 eV, which is attributed to vari- 431. ations of the domain-wall energy in the vicinity of [5] E. Mentz et al., Phys. Rev. B (1998), submitted for publica- Cu(1 0 0)-surface steps. tion. [6] F. Huang et al., Phys. Rev. B 49 (1994) 3962. [7] A. Kirilyuk et al., J. Magn. Magn. Mater. 159 (1996) Acknowledgements L27. [8] A. Bauer, E. Mentz, P. Jensen, and G. Kaindl, to be This work was supported by the Sfb-290 of the submitted. Deutsche Forschungsgemeinschaft. [9] E. Fatuzzo, Phys. Rev. 127 (1962) 1999.