Spin-flop transition driven by exchange springs in ErFe2/YFe2 multilayers K. N. Martin, K. Wang, G. J. Bowden and P. A. J. de Groot School of Physics and Astronomy, University of Southampton, SO17 1BJ, United Kingdom J. P. Zimmermann and H. Fangohr Computational Engineering and Design Group, School of Engineering Sciences, University of Southampton, SO17 1BJ, United Kingdom R.C.C. Ward Clarendon Laboratory, Oxford University, OX1 3PU, United Kingdom 1 Abstract. Magnetization loops for (110) ErFe2/YFe2 multilayer films grown by molecular beam epitaxy (MBE) are presented and discussed. The easy axis for the hard ErFe2 layers is near an out of plane <111> type crystal axis. At low temperatures there is just one irreversible switch of the hard layers, accompanied by the formation of magnetic exchange springs in the soft YFe2 layers. However, above a certain temperature the coercivity changes sign and there are additional high field transitions. This cross-over temperature, TCO, depends upon the composition of the multilayers. In sufficiently high fields, perpendicular to the multilayer film plane, the energy is minimized by an exchange spring driven multilayer spin-flop state. The composition dependence of TCO is explained with a simple energy argument. 2 There is currently a great deal of interest in exchange spring magnets, which have potential applications in data storage media1-3, permanent magnets4,5 and MEMS.6-8 REFe2 (RE=Rare earth) Laves phase films and multilayers grown by molecular beam epitaxy (MBE) have proved to be excellent model systems for the study of exchange spring phenomena.9 They display a rich variety of features, including exchange spring induced giant magnetoresistance.10 Additional features continue to be discovered. Xray Magnetic Circular Dichroism (XMCD)11 and polarised neutron reflectometry (PNR)12 have recently been used to study YFe2 dominated DyFe2/YFe2 systems. These techniques revealed complex reversal processes. Similarly complex magnetic re-orientations occur in ErFe2/YFe2 superlattices above a cross over temperature, TCO. In this paper we report on the multilayer composition dependence of TCO. The MBE procedure used to grow the ErFe2/YFe2 superlattices is described elsewhere.13 A 100 Å Nb buffer and 20 Å Fe seed layer were deposited onto epi- prepared 1120 sapphire substrate of dimensions 10 mm×12 mm. The Laves phase material was grown in (110) orientation, with the major axes parallel to those of Nb. This was achieved by co-deposition of elemental fluxes at a substrate temperature of 400°C. The resulting superlattices have good crystalline quality throughout. The magnetic data was obtained using a vibrating sample magnetometer, using applied fields Bapp of up to 12 T, within a temperature range of 10 to 310 K.14 In all cases the field was applied along the [110] growth-axis, perpendicular to the plane of the multilayer. The ErFe2 intermetallic compounds are characterised by (i) antiferromagnetic coupling between the Er and Fe moments and (ii) the strong (~600K) Fe-Fe exchange. 3 This leads to antiferromagnetic coupling between the net magnetisation of the hard (ErFe2) and soft (YFe2) layers in zero field. The easy axis of the hard layers is near an out of plane <111> type crystal axis. The multilayers discussed in this paper have relatively thick soft magnetic layers. At low temperatures the magnetic reversal of such samples is dominated by exchange springs in the soft layers. There is just one irreversible switch of the hard layers. However, above a cross-over temperature TCO significant changes occur. Micromagnetic modelling, performed using the finite difference method with the Object Oriented Micro-Magnetic Framework (OOMMF) software15, shows that these changes are due to different spin configurations that minimise the energy of the multilayer at high temperatures. In particular, in high fields and high temperatures the energy of the multilayer is minimised by an exchange spring driven spin-flop.16 In this paper we reveal the dependence of TCO on soft layer thickness. This composition dependence is explained by energy considerations. Figure 1. Magnetisation curves for (a) [ErFe2(50 Å)/YFe2(150 Å)] × 20 (SL1) and (b) [ErFe2(50 Å)/YFe2(100Å)] × 27 (SL2) at 100 K. The arrows indicate the direction in which the field is being swept. 4 Fig. 1 shows experimental magnetisation curves obtained for the multilayers [ErFe2(50 Å)/YFe2(150 Å)] × 20 (SL1) and [ErFe2(50 Å)/YFe2(100Å)] × 27 (SL2) at 100 K. Both magnetisation curves exhibit negative coercivity, a familiar feature of exchange-spring multilayers with thick soft magnetic layers.17 The magnetic reversal is dominated by exchange springs in the soft YFe2 layers. There is only one irreversible switch of the hard layers, beginning at ± 4.5 T and ± 3.9 T for SL1 and SL2 respectively. This behaviour is similar to that observed for DyFe2/YFe2 multilayers18, except that the spin configuration is now out of plane. Figure 2. Magnetisation curves for (a) (SL1) and (b) (SL2) at 150 K. The arrows indicate the direction in which the field is being swept. Fig. 2 shows magnetisation curves for the same superlattices at 150 K. The result for SL2 is similar to that at 100 K. The coercivity is again negative, the reversal is dominated by exchange springs and there is one irreversible switch of the hard layers starting at ± 3.5 T. However, significant changes have occurred in the magnetisation loop of SL1. The coercivity is now positive and there are additional irreversible 5 transitions at high fields. The irreversible nature of these transitions has been confirmed by measuring partial hysteresis loops. In high fields the magnetisation of the hard layers points close to an in-plane <111> crystal axis, while the average magnetisation of the soft layers is out of plane close to the applied field direction. The average magnetisation of the soft layers points close to the applied field direction. As the field is reduced to below 5 T the hard layer spins begin to switch to point out of plane, nearly opposite the applied field, while the soft layer magnetisation remains close to the applied field direction. This state remains until the field goes negative. But at the coercive field of 0.4 T, the hard layers switch to point out of plane again, nearly opposite to the reversed field direction, while the soft layer magnetisation switches to point primarily along the field direction. Finally, starting at -4.5 T the hard layer spins switch again to point along an in plane <111> axis. Thus the magnetisation loop is characterised by three irreversible switches, as the field is swept from a large positive value to a large negative value. Figure 3. Magnetisation curves for (a) (SL1) and (b) (SL2) at 200 K. The arrows indicate the direction in which the field is being swept. 6 Fig. 3 shows magnetisation curves for SL1 and SL2 at 200 K. The magnetic reversal of both multilayers now follows the exchange spring driven spin-flop mechanism described above for SL1. Starting at high fields, the first switch begins at 6 T (5 T) for SL1 (SL2). The coercivity is 0.15 T (0.46 T) and the third switch starts at 5 T (4 T). These results demonstrate the dependence of TCO on multilayer composition. TCO is 135(15) K for SL1 and 185(15) K for SL2. We interpret the experimental results as follows. The dependence of TCO on multilayer composition can be explained by energy considerations. The three main contributions to the energy are the anisotropy, exchange and Zeeman energy. At low temperatures the strong anisotropy of the hard layers dictates that the Er moments are aligned close to the applied field direction at high field. This in turn means that tight exchange springs must form in the soft layers to reduce the Zeeman energy, at a cost in exchange energy. Then, as the field is reduced, the exchange springs unwind reversibly. Finally, in a large enough reverse field the hard layers switch irreversibly to re-align with the applied magnetic field direction. This switch is accompanied by the formation of exchange springs in the soft layers. As the temperature is increased two main changes occur. The Er anisotropy associated with the hard layers falls off rapidly, accompanied by a decrease in the Er magnetic moment. Consequently, the soft YFe2 layers start to dominate the magnetic behaviour. At low fields the soft layer magnetisation points along the field direction, and the hard layer magnetisation points opposite the field. In a sufficiently high field the hard layer magnetisation switches irreversibly to point along an in plane <111> direction. This transition is driven by the gain in Zeeman energy of the hard layers, but at a cost of 7 anisotropy energy. It is accompanied by the formation of tight exchange springs in the soft layers. The strong Fe-Fe exchange energy prevents the hard layer magnetisation from simply aligning along the field direction. We call this spin configuration an exchange spring driven multilayer spin-flop. Thus TCO can be viewed as the temperature at which the soft layers begin to dominate the magnetic reversal of the superlattice. In summary, at temperatures below TCO, the magnetic reversal of YFe2 dominated superlattices is characterised by exchange springs. There is only one irreversible switch of the hard layers. The exchange spring driven spin flop occurs when the YFe2 layers dominate the magnetic response of the superlattice. So the thicker the soft layers are the lower the cross-over temperature TCO. Above this temperature the entire character of the hysteresis loop changes, with additional transitions at high fields. This is consistent with the experimental results for SL1 and SL2. 8 References. 1 D. Suess, T. Schrefl, R. Dittrich, M. Kirschner, F. Dorfbauer, G. Hrkac, and J. Fidler, J. Magn. Magn. Mater. 290, 551-554 (2005). 2 D. Suess, T. Schrefl, S. Fahler, M. Kirschner, G. Hrkac, F. Dorfbauer, and J. Fidler, Appl. Phys. Lett. 87, 12504 (2005). 3 R. H. Victora and X. Shen, IEEE Trans. Magn. 41, 537-542 (2005). 4 R. Skomski and J. M. D. Coey, IEEE Trans. Magn. 29, 2860-2862 (1993). 5 R. Skomski and J. M. D. Coey, Phys. Rev. B 48, 15812 (1993). 6 M. R. J. Gibbs, J. Magn. Magn. Mater. 290-291, 1298-1303 (2005). 7 A. Ludwig and E. Quandt, J. Appl. Phys. 87, 4691-4695 (2000). 8 C. T. Pan and S. C. Shen, J. Magn. Magn. Mater. 285, 422-432 (2005). 9 K. Dumesnil, M. Dutheil, C. Dufour, and P. Mangin, Phys. Rev. B 62, 11361140 (2000). 10 S. N. Gordeev, J.-M. L. Beaujour, G. J. Bowden, B. D. Rainford, P. A. J. de Groot, R. C. C. Ward, M. R. Wells, and A. G. M. Jansen, Phys. Rev. Lett. 87, 186808 (2001). 11 K. Dumesnil, C. Dufour, P. Mangin, F. Wilhelm, and A. Rogalev, J. Appl. Phys. 95, 6843-6845 (2004). 12 M. R. Fitzsimmons, S. Park, K. Dumesnil, C. Dufour, R. Pynn, J. A. Borchers, J. J. Rhyne, and P. Mangin, Phys. Rev. B 73, 134413 (2006). 13 V. Oderno, C. Dufour, K. Dumesnil, P. Mangin, and G. Marchal, J. Cryst. Growth 165 (1996). 14 M. Sawicki, G. J. Bowden, P. A. J. de Groot, B. D. Rainford, R. C. C. Ward, and M. R. Wells, J. Appl. Phys. 87, 6839-6841 (2000). 9 15 M. J. Donahue and D. G. Porter, OOMMF Users Guide (NIST, Gaithersburg, MD, 1999). 16 K. N. Martin, K. Wang, G. J. Bowden, A. A. Zhukov, P. A. J. de Groot, J. P. Zimmermann, H. Fangohr, and R. C. C. Ward, Appl. Phys. Lett. 89, 132511 (2006). 17 J.-M. L. Beaujour, S. N. Gordeev, G. J. Bowden, P. A. J. de Groot, B. D. Rainford, R. C. C. Ward, and M. R. Wells, Appl. Phys. Lett. 78, 964-966 (2001). 18 M. Sawicki, G. J. Bowden, P. A. J. de Groot, B. D. Rainford, J.-M. L. Beaujour, R. C. C. Ward, and M. R. Wells, Phys. Rev. B 62, 5817-5820 (2000). 10