Physica B 267}268 (1999) 207}210 Investigation of magnetic coupling phenomena in Fe \VCrV/Cr-superlattices with spin-polarized neutrons R. Siebrecht *, A. Schreyer , T. Schmitte , W. Schmidt , H. Zabel Institut fu(r Festko(rperphysik, Ruhr-Universita(t Bochum, 44780 Bochum, Germany Institut Laue}Langevin, 38042 Grenoble cedex 9, France NIST Center for Neutron Research, Gaithersburg, MD 20899, USA Abstract We present the results of temperature dependent measurements of magnetically coupled Fe \VCrV/Cr-superlattices. These results are supplementary to the ones known for non-collinearly coupled Fe/Cr-superlattices. By systematically varying the Cr concentration x we cover a wide range of the Fe \VCrV-phase diagram. As an experimental technique spin-polarized neutron re#ectivity with spin analysis and high-angle neutron scattering proves to be ideal for this work. 1999 Published by Elsevier Science B.V. All rights reserved. Keywords: Magnetic exchange coupling; Curie temperature; NeHel temperature; Neutron re#ectivity 1. Introduction wave in the bulk [4], should play an important role in explaining the magnetic coupling in the Fe/Cr- Twelve years ago an antiferromagnetic (AFM) system [5,6]. It was, however, only recently that by interlayer coupling of ferromagnetic (FM) thin use of high angle neutron scattering and polarized "lms separated by Cr spacer layers was reported neutron re#ectometry (PNR) experiments the NC [1]. Since then the Fe/Cr system has been of con- coupling in Fe/Cr-superlattices could be related siderable interest in the "eld of magnetic exchange with the spin structure of the Cr spacer layers [7,8] coupling in metallic superlattices. After the col- In agreement with the proximity magnetism model linear (C) AFM coupling, the non-collinear (NC) for NC coupled multilayers [9] it was demon- coupling was discovered [2,3]. It was pointed out strated that the NC coupling in Fe/Cr-superlattices that the magnetic structure of the Cr-spacer layer, originates from a frustrated spiral modulation of known to be an incommensurate AFM spin density the Cr moments [7]. Temperature dependent measurements showed that for the Cr NeHel temper- ature ¹, ! "500 K, the NC interlayer coupling vanished. To get a deeper insight into this phenom- * Correspondence address: Institut Laue}Langevin, 38042 Grenoble cedex 9, France. Tel.: #33-4-76-20-75-89; fax: #33-4- enon we changed those system parameters which 76-20-71-20. are of importance for the magnetic coupling. One E-mail address: siebrech@ill.fr (R. Siebrecht) way to do this is to exchange the Fe layers by an 0921-4526/99/$ } see front matter 1999 Published by Elsevier Science B.V. All rights reserved. PII: S 0 9 2 1 - 4 5 2 6 ( 9 9 ) 0 0 0 6 8 - X 208 R. Siebrecht et al. / Physica B 267}268 (1999) 207}210 Fe \VCrV alloy. This alloy is FM for a wide range combining the results of energy dispersive X-ray of x and its bulk phase diagram is well established analysis and superlattice periods , measured with [10]. In this way, the in#uence of the Fe \VCrV FM neutron re#ectivity. intralayer order on the Cr AFM intralayer order and vice versa can be studied systematically. With increasing x, both the average magnetic moment 3. Experimental per atom and, of even more interest, the Curie temperature ¹ Polarized neutron scattering techniques are ! $ ! , will be shifted to lower values. By varying ¹ ideally suited for the nondestructive study of the ! $ ! , the e!ect on ¹, ! can be studied. ¹ magnetic pro"le in thin metallic "lms [14}16] . By ! $ ! can be selected to be lower or even higher than ¹ means of PNR with spin analysis (SA) we deter- , ! . By varying the sample temper- ature for a given x one may switch o!a or ona the mined the in-plane vector magnetization pro"le coupling by either crossing ¹ [17,18] of the Fe ! $ ! or ¹, ! . \VCrV layers along the axis per- pendicular to the sample surface. was also deter- mined by neutron re#ectometry since the X-ray 2. Samples scattering contrast between the FeCr-alloy and the Cr was not su$ciently large to result in reasonable A set of [Cr/Fe X-ray re#ectivity spectra. With temperature depen- \VCrV],/Cr/Nb/Al O (1 1 0 2)- superlattices with di!erent x were prepared using dent PNR with SA, we obtained the transition molecular beam epitaxy. All samples were grown temperatures ¹! $ ! . All neutron re#ectivity ex- under the same conditions [7] since structural periments were carried out on the high #ux neutron properties may in#uence the coupling mechanism re#ectometer ADAM at the ILL [19]. A "rst [9,11}13]. Table 1 summarizes the samples and sample was studied with high angle neutron scat- their compositions. The Fe tering to access the spin structure of the Cr spacer \VCrV "lm thicknesses t layers on the triple axis spectrometer IN-12 at ILL. $ ! lie between 24 and 31 As. The Cr spacer layer thickness t A single purely magnetic peak of the commensurate ! was chosen to be smaller than 45 As, since in this case the Cr AF structure is known to be AF structure of the BCC-Cr is studied around commensurate below ¹ +0 0 1, positions [4,7]. , ! in Fe/Cr samples, fa- vouring magnetic coupling [7]. The large number of superlattice periods N*60 increases the 4. Results amount of Cr, which yields a better signal to back- ground ratio in neutron scattering experiments. The results obtained by means of PNR with SA X and the thicknesses t! , t$ ! were determined by are summarized in Table 1. No NC coupling was Table 1 List of [Cr/Fe \VCrV],/Cr/Nb/(1 1 0 2)Al O samples with systematic variation of the Cr concentration x. The samples were grown under equivalent conditions by molecular beam epitaxy. The spacer layer thicknesses, t! (45 As, was chosen to assure a commensurate AF spin structure. The large number of superlattice periods increases the amount of Cr which improves the signal to background ratio in neutron scattering experiments. The measured Curie temperatures for samples with higher concentrations are considerably lower than the bulk values x t! t$ ! Superlattice Coupling type Bulk T! $ ! (K) T! $ ! (K) (at% Cr) (As) (As) period N at ¹"20 K literature [10] measured 38 28 31 100 AFM 820 } 45 40 33 100 FM 760 770 57 28 24 60 AFM 490 450 66 36 27 190 FM 325 210 70 22 26 60 FM 250 185 80 21 25 60 (Spin glass) 55 } R. Siebrecht et al. / Physica B 267}268 (1999) 207}210 209 observed at low temperatures, ¹)20 K. This was three of the four re#ectivity cross R>>, R\\ and an unexpected result for the Fe \VCrV/Cr samples R>\"R\>. A clear chemical "rst-order Bragg suggesting that their interface topology is di!erent peak (R>>, R\\) and a purely magnetic half order to that of the Fe/Cr samples [7,12]. For x"0.8 no Bragg peak (R\>) at ¹"35 K are observed. This remanent magnetization of the FeVCr \V layer was result and the fact that there was no intensity split- found, consistent with spin glass behaviour. Bulk ting of R>> and R\\ at the "rst order peak, Fe}Cr alloys exhibit such spin glass character be- indicates an AFM alignment perpendicular to the tween 81 at% Cr and 84 at% Cr [10] below about neutron polarisation axis. While the temperature 25 K. Above this temperature the bulk alloy be- was increased, PNR radial scans with SA of both comes paramagnetic. A comparison of the mea- peaks were taken in order to record the temper- sured Curie temperatures with those of the bulk ature dependence of the magnetization vector shows that there is a considerable reduction of pro"le. The half order peak intensity in R\> ¹! $ ! for higher Cr concentrations. ¹! $ ! with diminished continuously, vanishing at ¹"380 K, x"0.38 was experimentally not accessible with whereas a splitting in the "rst order intensi- our cryo-furnace. Inferring from the continuous ties appeared as shown by the PNR data in the change in the transition temperatures, ¹! $ ! for inset of Fig. 1. Thus a rotational transition from this sample should be comparable to the bulk. The AFM towards FM takes place. Beyond this point AFM coupled samples show a change in the coup- there is only a decrease in the magnetic net moment ling type with increasing temperature. Fig. 1 shows up to ¹! $ ! "450 K. For the FM sample with Fig. 1. Re#ectivity measurement with spin-polarized neutrons and spin analysis at ¹"35 K for a Fe Cr /Cr sample. The half order Bragg peak in R>\("R\>) and no intensity splitting at the chemical "rst order Bragg peak position in R>> and R\\ indicate an AFM coupling perpendicular to the neutron polarisation axis. The high intensity in R>> and R\\ at the half order position is related to a high di!use magnetic scattering which was con"rmed with o! specular scans. The inset documents the magnetic reorientation from AFM to FM by its temperature-dependent splitting. For more details see text. 210 R. Siebrecht et al. / Physica B 267}268 (1999) 207}210 x"0.66, ¹! $ ! was found to be 180 K. A "rst References high angle neutron scattering experiment yields a Cr NeHel temperature ¹ [1] P. GruKnberg et al., Phys Rev. Lett. 57 (1986) 2442. , ! "360 K for this sample which is about 30% smaller than in equiva- [2] M. RuKhrig et al., Phys. Stat. Sol. (a) 125 (1991) 635. lent Fe/Cr-samples [7]. [3] B. Heinrich et al., Phys. Rev. B 44 (1991) 9348. [4] E. Fawcett, Rev. Mod. Phys. 60 (1988) 209. [5] J. Unguris, R.J. Cellotta, D.T. Pierce, Phys. Rev. Lett. 67 (1991) 140. 5. Conclusions [6] D. Stoe%er, F. Gautier, J. Magn. Magn. Mater. 121 (1993) 259. In a set of Fe [7] A. Schreyer, C.F. Majkrzak, Th. Zeidler, T. Schmitte, \VCrV/Cr-superlattices with a sys- P. BoKdeker, K. Theis-BroKhl, A. Abromeit, J.A. Dura, tematic variation of x we found reduced Curie tem- T. Watanabe, Phys. Rev. Lett. 79 (1997) 4914. peratures ¹! $ ! compared to bulk FeCr-alloys. [8] E.E. Fullerton, S.D. Bader, J.L. Robertson, Phys. Rev. Instead of an expected NC coupling all the samples Lett. 77 (1996) 1382. couple either FM or AFM. For the AFM samples [9] J.C. Slonczewski, J. Magn. Magn. Mater. 150 (1995) 13. a continuous magnetic reorientation into the FM [10] Landolt BoKrnstein, New Series III/19a, 1986, p. 338, 342. [11] J.A. Wolf, Q. Leng, R. Schreiber, P.A. GruKnberg, W. Zinn, state takes place with increasing temperature. ¹, ! , J. Magn. Magn. Mater. 121 (1993) 253. which we have determined for one sample so far, is [12] A. Schreyer et al., Europhys. Lett. 32 (1995) 595. also considerably lower than in the case of Fe/Cr- [13] A. Schreyer et al., Europhys. Phys. Rev. B 52 (1995) superlattices. Thus, ¹ 16066. , ! can be varied via ¹! $ ! , indicating a proximity e!ect between the FM [14] C.F. Majkrzak, J.W. Cable, J. Kwo, M. Hong, D.B. McWhan, Y. Yafet, J.V. Waszczak, C. Vettier, Phys. Rev. FeVCr \V and the AF Cr layers. We are currently Lett. 56 (1986) 2700. investigating this proximity e!ect systematically as [15] S. J Blundell, J.A.C. Bland, Phys. Rev. B 46 (1992) a function of alloy concentration x. 3391. [16] H. Zabel, Appl. Phys. A 58 (1994) 159. [17] G.P. Felcher, R.O. Hilleke, R.K. Crawford, J. Hauman, R. Kleb, G. Ostrowski, Rev. Sci. Instrum. 58 (1987) Acknowledgements 609. [18] N.K. Pleshanov, Z. Phys. B 94 (1994) 233. We gratefully acknowledge the hospitality of the [19] R. Siebrecht, A. Schreyer, U. Englisch, U. Pietsch, H. ILL as well as "nancial support by the German Zabel, Physica B 241}243 (1998) 169. BMBF (03-ZA4BC2-3) and DFG via SFB 166.