Journal of Magnetism and Magnetic Materials 198}199 (1999) 387} 390 Symmetry in#uence on interlayer coupling in epitaxial Co/Cr trilayers grown on MgO (1 0 0) and (1 1 0) substrates J. Zachary Hilt , J. Johanna Picconatto , Alexandra O'Brien , Michael J. Pechan *, Eric E. Fullerton Department of Physics, Miami University, Oxford, OH 45056, USA Materials Science Division, Argonne National Laboratory, Argonne, IL 60439, USA Abstract Trilayers of Co(15 As)/Cr(x)/Co(17 As) have been epitaxially sputtered onto MgO (1 0 0) and (1 1 0) substrates coated with Cr(1 0 0) and (2 1 1) bu!er layers, respectively. The Cr thickness x is varied from 6 to 50 As. Both sample sets have the Co c-axis lying in the plane of the "lm, however, due to the four-fold symmetry of MgO (1 0 0), the Co layers form bicrystals with perpendicularly oriented c-axis. Ferromagnetic resonance measurements clearly show large two-fold and four-fold magnetic in-plane anisotropy for MgO (1 1 0) and MgO (1 0 0) samples, respectively. Magnetization measure- ments reveal interlayer coupling strengths peaked at a Cr thickness of approximately 10 As for both symmetries, but the strength decreases more slowly with increasing t! in the MgO(1 0 0) system. 1999 Elsevier Science B.V. All rights reserved. Keywords: Trilayers; FMR spectra 1. Introduction from 15 to 100 As. This present study on structurally simpler trilayers has been undertaken to address the Recently, investigators have utilized epitaxial growth in#uence of the sample symmetry on the anisotropy and techniques to deposit coherent HCP Co layers with in- interlayer exchange coupling. plane c-axis orientation [1}5]. A series of investigations have been reported [6}8] on the anisotropy and ex- change coupling in Co/Cr superlattices, in which the 2. Experimental details in-plane symmetry was primarily two-fold, although for some samples with lower Co thicknesses four-fold sym- Two series of Co(15 As)/Cr(x)/Co(17 As) trilayers with metry was observed. An investigation of Co/Cr multi- (6) t layer samples grown on MgO (1 1 0) and MgO (1 0 0) ! )50 As) were epitaxially sputtered onto single crystal MgO (1 0 0) and (1 1 0) substrates. The substrates was reported [9] in which the sample symmetry was were mounted side by side onto the sample holder and determined by the substrate symmetry. In particular, the simultaneously deposited. A 100 As Cr layer was intially four-fold symmetry was realized for Co thickness ranging deposited at a substrate temperature of 6003C resulting in epitaxial Cr (2 1 1) and (1 0 0) bu!er layers on MgO (1 1 0) and (1 0 0), respectively [10]. The substrate was * Corresponding author. Tel.: 513-529-4518; fax: 513-529- then cooled to 1503C and the trilayers were grown by 5629. sequential deposition of the Co and Cr layers. This E-mail address: pechanmj@muohio.edu (M.J. Pechan) growth procedure has been successfully used to grow Present address: IBM Almaden Research Center, San Jose, Fe/Cr and Co/Cr superlattices [2,10]. Due to the limited CA 95120 material in these trilayer samples, X-ray measurements 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 1 4 1 - X 388 J. Zachary Hilt et al. / Journal of Magnetism and Magnetic Materials 198}199 (1999) 387}390 are not feasible. However the trilayers were grown under identical conditions used to prepare [Co (20 As)/ Cr(7)x)22 As)] multilayer samples [9]. Therefore it is assumed that the trilayers possess structure similar to that found for the multilayers. Those results indicate that BCC-Cr (2 1 1) and coherent HCP-Co (1 1 0 0) are formed on MgO (1 1 0) substrates and that BCC-Cr (1 0 0) and epitaxial, strained HCP-Co (1 1 2 0 are formed on MgO (1 0 0). In both cases, the Co grows Fig. 1. Ferromagnetic resonance "elds as a function of in-plane with the c-axis in plane of the "lm, but the MgO (1 0 0) angle. Symbols are data, solid line is "t to model described in the series has two equivalent perpendicular directions in text. which the c-axis may be oriented [11]. In-plane anisotropy symmetry was determined using 35GHz ferromagnetic resonance (FMR) with the sample arise from surface anisotropy or stress-induced crystal placed "lm side down on the bottom of TE102 mode "elds normal to the "lm surface. All MgO (1 1 0) samples cavity and the magnet rotated about the sample. Hyster- reported here have anisotropy results similar to the esis loops were measured on a vibrating sample mag- above. netometer (VSM) built in our lab primarily by the two As mentioned above, the MgO (1 0 0) substrate pro- undergraduate authors (JJP and JZH). Absolute magne- motes bicrystalline Co growth with mutually orthogonal tization values were obtained on a SQUID mag- c-axis leading to four-fold in-plane anisotropy. As has netometer. been pointed out previously [9], in the limit of strong inter-crystallite coupling, the Co layer behaves as a single moment with anisotropy given by the sum of E( ) and 3. Results and discussion E( # /2) in Eq. (1), resulting in an average in-plane anisotropy determined by Saturation magnetization values for the Co in the samples reported vary in the range of 1100 to E( )" K cos (2 ). (2) 1200 emu/cm . While less than the bulk value of 1400 emu/cm , these results are consistent with values Again very good agreement between experiment and reported by others [6] and are consistent with a small theory is observed as the model yields a value of amount of alloying at the Co/Cr interface. K "9.9($0.9);10 erg/cm for the t! "10 As MgO FMR spectra for all samples reported are symmetric, (1 0 0) sample. This is signi"cantly larger than for the single mode with linewidth of approximately 1000 G and two-fold samples, being still only slightly less than 70% signal-to-noise ratio exceeding 100. The two-fold sym- the K for bulk Co. All MgO (1 0 0) trilayer samples metry of the MgO (1 1 0) series of samples and the four- reported here have anisotropy results similar to the fold of the (1 0 0) series is readily apparent in the FMR above. resonance positions as exemplifed in Fig. 1. These reso- In none of the FMR spectra observed is there any nance positions "t very well a model that assumes the evidence for modes other than a single uniform reso- standard form for Co uniaxial anisotropy nance mode. This supports, in the case of the four-fold samples, the assumption of strong inter crystalline cou- E( )"K pling within a Co layer (if the crystallites were weakly sin #K sin , (1) coupled, their magnetizations would precess indepen- where is the angle between the magnetization M and dently and two resonances, one from each c-axis orienta- the Co c-axis, and K and K are "rst-and second-order tion, would be observed at each in-plane angle). Such anisotropy energy densities. The model also includes an weak intercrystalline coupling was reported in one Co/Cr out-of-plane anisotropy term which is cast in the form of multilayer sample [8]. The lack of additional modes also a shape anisotropy 4 M . The two-fold sample means that we were not observing the anticipated out- shown in Fig. 1 is characterized by K "1.8($0.4) of-phase resonance modes associated with interlayer ;10 erg/cm , K "0.55($0.12); 10 erg/cm . These coupling e!ects. These out-of-phase modes may be sup- K values are approximately 40% of the bulk Co values, pressed somewhat by the large in-plane anisotropies but have the same 3 to 1 ratio as is observed in the bulk. present in the samples. The out-of-plane anisotropy for the two-fold sample Although we did not observe the coupling e!ects in shown in Fig. 1 is characterized by M " M , which FMR spectra, anitiferromagnetic (AF) interlayer cou- implies an easy-axis out-of-plane anisotropy that pling is clearly manifest as symmetrically o!set hysteresis counters the easy-plane demagnetization. This rather loops shown in Figs. 2 and 3, where, in all cases, the "eld large additional (4 M !4 M ) anisotropy could is applied along the in-plane easy axis. The interlayer J. Zachary Hilt et al. / Journal of Magnetism and Magnetic Materials 198}199 (1999) 387}390 389 Fig. 4. Exchange energy densities as a function of t! . Diamonds and squares are MgO (1 1 0) and (1 0 0), respectively. Fig. 2. Hysteresis loops for MgO (1 1 0) sample series. Magnetic "eld as applied in-plane along the easy axis. coupling peak occurs in the 8}10 As range whereas the peak occurred at 13 As in similarly prepared superlattice samples [9]. As seen in Fig. 3, where again the "eld is applied along one of the in-plane easy axes, the MgO (1 0 0) system also exhibits o!set hysteresis loops indicative of antiferromag- netic interlayer coupling at t! "8, 10, 11.5 and 13 As. The approach to saturation for the four-fold system is in principle more complicated than for the two-fold, often requiring "tting via energy minimization to reliably ex- tract coupling energies. However, in the case where the anisotropy is considerably larger than the coupling en- Fig. 3. Hysteresis loops for MgO (1 0 0) sample series. Magnetic ergy one can demonstrate that the centroid of the o!set "eld is applied in-plane along one of the easy axes. hysteresis is the e!ective coupling "eld. For example, minimization simulations were done utilizing Zeeman, coupling and fourfold anisotropy energies (as per Eqs. (2) coupling energy per unit area may be simply expressed as and (3)) with the "eld along an easy in-plane axis. With (ignoring biquadratic coupling) anisotropy "elds 2, 4 and 10 times the exchange "eld, switching "elds were 10, 5 and 0% greater, respectively, Et! "!J cos( ! ), (3) than the exchange "eld. In our MgO (1 0 0) AF coupled where J'0 ((0) implies ferromagnetic (antiferromag- samples, the anisotropy "elds (approximately 3500 G) netic) coupling, t are 3}5 times greater than the switching "elds. Given that ! is the Co layer thickness and repre- sent the angle between M our accuracy in determining the switching centroid is no (one for each Co layer) and the "eld direction which is also the in-plane easy direction. better than$5%, we conclude that J may be obtained by In the MgO (1 1 0) system, samples with t Eq. (4) above. The results are plotted in Fig. 4. Even ! "8, 10, 11.5 and 13 A though both occur via domain wall motion, the switching s show an abrupt transition from antiparal- lel to parallel alignment. Given the strong in-plane an- transitions are not as sharp in the (1 0 0) as in the (1 1 0). isotropy in these samples, the antiparallel magnetizations This is likely due to impeded wall motion in the (1 0 0) are aligned along the easy axis ( due to domain wall pinning arising from the bicrystal "0, " ). As switches from to 0 (via domain wall propagation) the an- nature of a Co layer. isotropy energy remains unchanged and therefore the Biquadratic coupling would not a!ect the (1 1 0) loops H seen in Fig. 2, however, if present, it should manifest itself is a direct measure of the e!ective coupling "eld. Since this transition occurs in a metamagnetic or non-equilib- as additional, low "eld, abrupt transitions in the (1 0 0) rium fashion, the Zeeman energy at the switching "eld is loops of Fig. 3. Since no such transitions are apparent, it equal to the change in coupling energy as must be concluded that any biquadratic coupling energy switches form to 0. Therefore, the magnitude of J is given by is negligible in these samples. The long-period oscillatory coupling (the period, J"M phase, and strength) in Fe/Cr has been shown to be t! H (4) independent of crystallographic orientation and, there- and the results are plotted in Fig. 4. The J values are fore, is believed to arise from a common feature of the Cr somewhat smaller over the same Cr thickness range than Fermi surface [10]. This isotropic behavior has been the Fe/Cr system [10]. However, the present J values are attributed to spanning vectors across a d-derived lens comparable to those observed in Co/Cr multilayers with [12] or the N-centered ellipse of the Cr Fermi surface Co c-axis in-plane [7]. It is interesting to note that the [13]. As seen in Fig. 4, the Co/Cr trilayer coupling 390 J. Zachary Hilt et al. / Journal of Magnetism and Magnetic Materials 198}199 (1999) 387}390 energies do depend on the Cr spacer-layer symmetries. References The (1 0 0) coupling is somewhat greater than that for the (2 1 1) spacers and persists through greater Cr thick- [1] A. Nakamura, M. Futamoto, Jpn. J. Appl. Phys. 32 (1993) nesses. These di!erences can be explained as a simple 1410. phase shift in the peak coupling strength as a function of [2] G.R. Harp, S.S.P. Parkin, Appl. Phys. Lett. 65 (1994) Cr thickness. Although the period of the exchange coup- 3063. ling depends predominantly on the details of the Cr [3] N. Metoki, W. Donner, H. Zabel, Phys. Rev. B 49 (1994) Fermi surface, the phase and strength of the coupling 17 351. [4] J.C.A. Huang, Y. Liou, Y.D. Yao et al., Phys. Rev. B 52 also depend on the overlap of the Co and Cr wavefunc- (1995) R13110. tions [13], and on the structure of the interface. Even [5] M. Grimsditch, E.E. Fullerton, R.L. Stamps, Phys. Rev. though the Co/Cr and Fe/Cr systems have the same Cr B 56 (1997) 2617. epitaxy, the magnetic layers have very di!erent symmet- [6] Th. Zeidler, F. Schreiber, H. Zabel et al., Phys. Rev. B 53 ries and, in addition, there is signi"cant dependence of (1996) 3256. the interfacial strains on crystallographic orientation in [7] Th. Zeidler, F. Schreiber, H. Zabel, J. Appl. Phys. 79 (1996) the Co/Cr system that is not common to the Fe/Cr 4793. system. [8] F. Schreiber, Z. Frait, Th. Zeidler et al., Phys. Rev. B 51 (1995) 2920. [9] J.J. Picconatto, M.J. Pechan, E.E. Fullerton, J. Appl. Phys. Acknowledgements 81 (1997) 5058. [10] E.E. Fullerton, M.J. Conover, J.E. Mattson et al., Phys. Rev. B 48 (1993) 15 755. This work was supported by US Department of En- [11] J.C.A. Huang, F.C. Tang, W.W. Fang et al., J. Appl. Phys. ergy, Basic Energy Sciences } Materials Sciences under 79 (1996) 4790. Contracts No. W-31-109-ENG-38 (ANL) and DE- [12] D.D. Koelling, Phys. Rev. B 50 (1994) 273. FG02-86ER45281 (Miami). [13] M.D. Stiles, Phys. Rev. B 54 (1997) 14 679.