Journal of Magnetism and Magnetic Materials 223 (2001) 299}303 Oscillatory exchange bias in Fe/Cr double superlattices L. Lazar , J.S. Jiang *, G.P. Felcher , A. Inomata , S.D. Bader Institut fu(r Experimentalphysik/Festko(rperphysik, Ruhr-Universita(t Bochum, 44780 Bochum, Germany Argonne National Laboratory, 9700 S. Case Ave. Argonne, IL 60439, USA Received 17 May 2000; received in revised form 10 October 2000 Abstract In the [Fe/Cr]$/CrV/[Fe/Cr]$ double superlattices consisting of a ferromagnetic Fe/Cr superlattice on top of an antiferromagnetic Fe/Cr superlattice, the exchange coupling between the superlattices is determined by the thicknesses (x) of the Cr spacer layer. The oscillating behavior of the exchange bias "eld of a series of (2 1 1)-oriented Fe/Cr double superlattices was determined by superconducting quantum interference device (SQUID) and magneto-optic Kerr e!ect (MOKE) measurements. For x'13 As a negative strongly oscillating character of the exchange bias was observed. At very thick x the exchange bias vanishes. The most immediate result is the fact that the exchange bias "eld is always negative, regardless of the sign of the coupling between the ferromagnetic and the antiferromagnetic superlattices. The detailed dependence of the exchange bias "eld as a function of the intersuperlattice thickness of Cr is explained in terms of the interaction between the two superlattices in collinear con"guration. 2001 Elsevier Science B.V. All rights reserved. Keywords: Exchange bias; Superlattice; Interlayer coupling; Exchange coupling; Epitaxy Exchange anisotropy is an e!ect caused by the applications of this e!ect make its study of even magnetic interface interaction between a ferromag- greater interest. The exchange bias e!ect is utilized net (F) and an antiferromagnet (AF) [1]. The fer- for permanent magnet materials [3], high-density romagnet magnetization (M}H) loop shift away recording media [4] and domain-stabilized record- from H"0 indicates the existence of an exchange ing heads [5]. anisotropy. The magnitude of this shift is know as In order to better understand the fundamental the exchange bias "eld H#. H# depends stronly on aspects of exchange bias we have studied the shift of many factors, including the spin structure at the the hysteresis loops in double superlattices, which interface, the antiferromagnetic anisotropy, rough- are arti"cial magnetic systems where the exchange ness, crystallinity, etc. [2]. The detailed dependence bias can be realized and analyzed with minimal from these factors has been the subject of an enor- material-related complexities [6]. The working mous experimental and theoretical endeavor dur- of these novel systems are constructed based ing the past years. The important technological on the well-established oscillatory interlayer exchange coupling in Fe/Cr so that the ferromag- * Corresponding author. Tel.: #1-630-252-4907; fax: #1- netic superlattice on top of the system is coupled 630-252-9595. ferromagnetically or antiferromagnetically to the E-mail address: jiang@anl.gov (J.S. Jiang). antiferromagnetic superlattice multilayer at the 0304-8853/01/$ - see front matter 2001 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 4 - 8 8 5 3 ( 0 0 ) 0 1 3 6 6 - 4 300 L. Lazar et al. / Journal of Magnetism and Magnetic Materials 223 (2001) 299}303 bottom. The Cr spacer layer thickness between the and F magnetization vectors are collinear, two superlattices is varied to provide a controlled H exchange coupling between the AF and F superlat- #J!"J ". This is why in almost all cases, in which the exchange bias has been experimentally tices. observed in AF/F pairs, this turned out to be nega- The double superlattice structure has many ad- tive [1,3,5]. However, H vantages over conventional systems for the study of #'0 in certain FeF/Fe [8] and MnF the exchange bias e!ect. Since the exchange bias /Fe [9] bilayers when the systems are cooled through the NeHel temperature of the AF e!ect is primarily an interface phenomenon, the in a large positive "eld. Positive exchange bias has "rst requirement for a systematic investigation is to been attributed to competition between the antifer- have highly ideal AF/F interfaces that are easy to romagnetic coupling at the AF/F interface and the characterize and to manipulate. This is satis"ed in Zeeman energy of the AF surface spins [9]. Up to the double superlattices: the long period of 18 As for now it was assumed that the coupling J interlayer coupling in Fe/Cr makes the double-  is the weak link, being smaller than both the interlayer superlattice structure less sensitive to roughness coupling in the F superlattice, J and amenable to characterization techniques such $, as well as the similar quantity for the AF superlattice, J as polarized nuetron re#ectivity measurement. $. Sup- pose that J A second prerequisite for exchange bias is magnetic  is considerably stronger than J$. Then, upon switching of the "eld, the magnetic anisotropy in the AF region. For double-superla- con"guration that is energetically most favorable is ttice systems this may be obtained as a growth- one in which a domain wall (or partial domain wall) induced uniaxial in-plane magnetic anisotropy. is created in the ferromagnet or the antiferromag- Fe/Cr (2 1 1) superlattices can be epitaxially sput- net [10}12]. Also, these hypothetical con"gura- tered on single-crystal MgO (1 1 0) substrates, with tions have to be compared with a scenario, in which the easy axis along the Fe/Cr [0 1 1] direction [7]. switching takes place with the creation of lateral In this paper we present the behavior of the ex- domains [13]. With the double-superlattice struc- change bias "eld in Fe/Cr(2 1 1) double superlatti- ture, we can simply vary the Cr spacer thickness to ces as a function of the thickness x of the Cr spacer vary J layer between the two superlattices, and relate it to , and to explore the interplay between J , J the exchange coupling between the two superlatti- $, J$ and the Zeeman energy. The double superlattices were grown via DC ces. magnetron sputtering onto single-crystal MgO At "rst thought the exchange "eld H# is strictly (1 1 0) substrates. The samples had a layer sequence proportional to J , the exchange energy between Cr the two superlattices. In Fig. 1 is shown that,  s/[Cr s/Fe s]/CrV/[Fe s/Cr s]/ Cr at least for a magnetic con"guration where AF  s /MgO(1 1 0). The 200 As Cr bu!er layer de- posited at 4003C established good epitaxy with the Fig. 1. Sketch representing the spin con"gurations of a double superlattice before (H'# , is a small but "nite value) and after (H(!"J "/N$t$M$) switching. The left panel is for ferromagnetic inter-superlattice coupling (J '0) and the right panel is for antiferromagnetic inter-superlattice coupling (J (0). In both cases, the exchange bias "eld is negative because the top Fe layer in the AF superlattice adopts a direction compatible with the sign of J . L. Lazar et al. / Journal of Magnetism and Magnetic Materials 223 (2001) 299}303 301 substrate. The double superlattice structure was grown at 903C and covered on top by a 50 As Cr layer. The thickness of the Cr spacer layer between the F and AF superlattices, x, was varied from 11 to 70 As to cover four oscillations of the sign of the inter-superlattice coupling. Care was taken to en- sure that the samples within the series are otherwise identical. Structural characterization of all pre- pared samples was done by small and high angle X-ray di!raction using Cu K? radiation. Even at large Cr spacer layer thicknesses the low-angle X- ray di!raction peaks were clearly resolved, indicat- ing that the "lms are well-layered. The Bragg peaks have a FWHM of 0.37}0.583 which leads to coher- ence lengths of 150}234 As. The typical values for layer roughness are shown in Ref. [14]. The magnetization measurements of all prepared samples were obtained with a superconducting quantum interference device (SQUID) mag- netometer and a magneto-optic Kerr e!ect (MOKE) system. The measurements were taken at room temperature and with the magnetic "eld ap- plied along the easy axis. With the SQUID mag- netometer we measured both full and minor hysteresis loops. The bias e!ect in the double super- lattice is obtained by aligning the magnetization of both F and AF superlattices in a high "eld of 15 kOe and then run the minor hysteresis loop. The shifted hysteresis loop is the signature of the un- idirectional anisotropy. Figs. 2(b) and (c) provide an example of the minor loops of two double super- lattices with (a) x"16 As, H#"!12Oe, and (b) x"20 As, H#"!38Oe. In all cases, the minor loops were characterized by very square hysteresis loops of relatively modest width. The value of the exchange bias "eld H# was obtained by taking the Fig. 2. Magnetization minor loops for double superlattices value of the magnetic "eld at mid-point of the loop. with: (a) x"13 As and H Fig. 3 gives the dependence of the exchange bias #"0 Oe; (b) x"16 As, H#"!12 Oe; (c) x"20 As, H#"!38Oe. "eld on the Cr spacer layer thicknesses x. One can see a strong oscillatory character for thin Cr spacer layer thicknesses, consisting of minima and maxi- concluded that the antiferromagnetic structure was ma, and the "nal damping of this oscillations at collinear to the ferromagnetic structure. It was also large Cr thicknesses. The solid line is guide for the con"rmed that in the two states of the minor loop eyes. The circles are representing the ferromagneti- the AF superlattice was "xed and only the spins of cally coupled interfaces and the squares are repres- the F superlattice were switched. The magnitude of enting the antiferromagnetically coupled interfaces. the exchange bias "eld was found to be equal to the On the basis of detailed neutron re#ectivity value expected of the exchange interaction between measurements [6,14] on one of the sample it was collinear AF and F superlattices [6]. Thus, it was 302 L. Lazar et al. / Journal of Magnetism and Magnetic Materials 223 (2001) 299}303 Fig. 3. The exchange bias "eld H# as a function of the Cr spacer layer thicknesses, x, for the analyzed double superlattices. The solid line is guide to the eyes. The circles are representing the F-coupled interfaces and the squares are representing the AF-coupled interfaces. justi"ed to use the classic formula for the magni- ercivity in conventional exchange biased systems tude of the exchange bias "eld as applied to systems has been associated with instability of antifer- of collinear spin structures H#"!"J "/N$t$M$, romagnetic grains in recent theories [16,17]. Our where N$ is the number of Fe layers in the fer- observation in double superlattices with very thin romagnetic superlattice, and t$ and M$ are the Cr spacers is consistent with them. thickness and saturation magnetization of the Fe In conclusion, we report the observation of an layers. The obtained oscillatory behavior of oscillatory behavior of the exchange bias "eld de- the exchange bias "eld versus the Cr spacer layer pending on the Cr spacer layer thicknesses between thicknesses follows the oscillatory interlayer coup- an antiferromagnetic and a ferromagnetic superla- ling in Fe/Cr (2 1 1) superlattices presented in ttice in novel double superlattice structures. The Ref. [7]. However, since the double superlattices do oscillatory behavior is similar to the oscillatory not require cooling in a "eld to establish exchange dependence of the coupling strength in Fe/Cr (2 1 1) bias, the top Fe layer in the AF superlattice adopts superlattices, which is a clear indication of a linear a direction compatible with the inter-superlattice dependence of the exchange bias "eld on the inter- coupling as illustrated in Fig. 1. Therefore, H# is facial exchange energy as it was expected. A strik- always negative regardless of whether the inter-super- ing e!ect is the observation of exchange bias "elds lattice coupling is ferromagnetic or antiferromagnetic. which is independent of the sign of the coupling. At thin Cr spacer layer thicknesses (x"13 As) the With double superlattices the AF/F coupling can experimental value of H# is zero within experi- be controlled by varying the spacer thickness as mental error. However, the MOKE hysteresis loop demonstrated. This ability to "ne tune the bias "eld represented in Fig. 2(a) is very wide, with coercivity could have implications on applications. H "42Oe. The value of H is comparable to the calculated value for H# using the classical formula for collinear spin structures mentioned above Acknowledgements (H#"94.1Oe). This suggests that the AF superla- ttice has switched irreversibly during the minor L. Lazar thanks the Deutsche Forschun- loop measurement [15]. The enhancement of co- gsgemeinschaft (DFG) (project ZA 161/16-1) for the L. Lazar et al. / Journal of Magnetism and Magnetic Materials 223 (2001) 299}303 303 "nancial support during the stay at the Argonne [7] E.E. Fullerton, M.J. Conover, J.E. Mattson, C.H. Sowers, National Laboratory. Work at Argonne was sup- S. Bader, Phys. Rev. B 48 (1993) 15755. ported by the US Department of Energy, O$ce of [8] J. NogueHs, D. Lederman, T.J. Moran, I.K. Schuller, Phys. Science, under contract No. W-31-109-ENG-38. Rev. Lett. 76 (1996) 4524. [9] C. Leighton, J. NogueHs, H. Suhl, I.K. Schuller, Phys. Rev. B 60 (1999) 12837. [10] D. Mauri, H.C. Siegmann, P.S. Bagus, E. Kay, J. Appl. References Phys. 62 (1987) 3047. [11] N.C. Koon, Phys. Rev. Lett. 78 (1997) 4865. [1] W.H. Meiklejohn, J. Appl. Phys. 33 (1962) 1328. [12] M. Kiwi, J. MejmHa-LoHpez, R.D. Portugal, R. RammHrez, [2] J. NogueHs, I.K. Schuller, J. Magn. Magn. Mater. 192 (1999) Europhys. Lett. 48 (1999) 573. 203, and references therein. [13] A.P. Malozemo!, Phys. Rev. B 35 (1987) 3679. [3] E.F. Kneller, R. Hawig, IEEE Trans. Magn. 27 (1991) [14] S.G.E. te Velthuis, G.P. Felcher, J.S. Jiang, A. Inomata, 3588. C.S. Nelson, A. Berger, S.D. Bader, Appl. Phys. Lett. 75 [4] M. Ohkoshi, K. Tamari, M. Harada, S. Honda, T. Kusuda, (1999) 4174. IEEE Trans. J. Magn. Japan 1 (1985) 37. [15] S.G.E. te Velthuis, J.S. Jiang, G.P. Felcher, Appl. Phys. [5] C. Tang, J. Appl. Phys. 55 (1984) 2226. Lett. 77 (2000) 2222. [6] J.S. Jiang, G.P. Felcher, A. Inomata, R. Goyette, C. Nel- [16] T.C. Schulthess, W.H. Bulter, Phys. Rev. Lett. 81 (1998) son, S.D. Bader, Phys. Rev. B 61 (2000) 9653. 4516. [17] M.D. Stiles, R.D. McMichael, Phys. Rev. B 59 (1999) 3722.