ARTICLE IN PRESS Journal of Magnetism and Magnetic Materials 298 (2006) 43–47 www.elsevier.com/locate/jmmm Coexistence of positive and negative exchange bias in CrMn/Co bilayers Nguyen Nguyen Phuoca,b,Ã, Nguyen Phu Thuya,c, Nguyen Anh Tuana, Le Thanh Hunga, Nguyen Trung Thanha, Nguyen Thanh Nama a International Training Institute for Materials Science, Hanoi University of Technology, Hanoi, Vietnam b Information Storage Materials Laboratory, Toyota Technological Institute, Nagoya, Japan c College of Technology, Vietnam National University, Hanoi, Vietnam Received January 2005; received in revised form 19 February 2005 Available online 24 March 2005 Abstract Exchange-biased CrMn/Co bilayers with various thicknesses of Co sputtered onto Si(1 0) substrates by the RF sputtering system have been studied Double-shifted loops have been observed with the thickness of Co layer in a narrow range and become single-shifted loops after some cycles of measurement Those results are interpreted as the association of positive and negative exchange bias r 2005 Elsevier B.V All rights reserved PACS: 75.70.Cn; 75.70.Ài; 75.25.+z; 75.30.Gw Keywords: Exchange bias; Magnetic thin film; Double-shifted loop; Training effect Introduction Discovered in 1956 [1], the phenomenon of exchange bias between an antiferromagnet (AF) and ferromagnet (FM) is of great interest due to its widespread application in spin valves and magÃCorresponding author Information Storage Materials Laboratory, Toyota Technological Institute, 2-12-1 Hisakata, Tempaku, Nagoya 468-8511, Japan Tel.: +81 52 809 1872; fax: +81 52 809 1874 E-mail address: nnguyenphuoc@yahoo.com (N.N Phuoc) netic tunnel junctions Nevertheless, its physical origin remains unanswered [2] Usually, exchange bias is described as an additional unidirectional anisotropy induced by the AF into the FM via exchange coupling at the interface, producing a single magnetic hysteresis loop shifted along the magnetic field axis after field cooling procedure through the Ne´el point of the AF The magnitude of this shift is named exchange bias field (HE) and in almost all cases, the magnetic hysteresis loop is shifted in the negative field if one defines the direction of the cooling field (HFC) as 0304-8853/$ - see front matter r 2005 Elsevier B.V All rights reserved doi:10.1016/j.jmmm.2005.03.006 ARTICLE IN PRESS N.N Phuoc et al / Journal of Magnetism and Magnetic Materials 298 (2006) 43–47 44 applied magnetic field This state manifests itself as a double hysteresis loop In this paper, we report the observation of the double-shifted loop in CrMn/Co bilayers with a proper thickness of Co layer A training-like effect has been observed in the sample exhibiting the double-shifted hysteresis loops, which becomes a single-shifted hysteresis loop after some cycles of measurement The results are interpreted as the association of positive and negative exchange bias, in which the portion of positive exchange bias 0.00030 0.002 0.00015 0.001 M (emu) M (emu) the positive direction This case is referred to as negative exchange bias The phenomenon of positive exchange bias was first observed in 1996 by Nogue´s et al [3] when studying the systems of Fe/FeF2 and Fe/MnF2 They found that the sign of exchange bias field changes from negative to positive as the cooling field increases Very recently, Roshchin et al [4] have found that the state of coexistence of positive and negative exchange bias can be achieved by cooling the sample of FeF2/Co in a properly chosen constant 0.00000 -0.00015 -0.001 tCo = 18 nm H (Oe) 1000 -400 2000 0.002 0.0005 0.001 M (emu) M (emu) -1000 0.0010 0.0000 tCo = 36 nm -1000 H (Oe) 1000 400 0.000 0.00150 0.0030 0.00075 0.0015 0.00000 -0.00075 tCo = 90 nm -400 2000 M (emu) M (emu) 200 H (Oe) -0.002 -0.0010 -200 200 H (Oe) 400 0.0000 -0.0015 tCo = 54 nm tCo = 108 nm -0.0030 -0.00150 -1000 -200 -0.001 -0.0005 -2000 tCo = 72 nm -0.002 -0.00030 -2000 0.000 -500 H (Oe) 500 1000 -400 -200 200 400 H (Oe) Fig Hysteresis loops of the CrMn(12 nm)/Co (x nm) samples measured at 123 K The thickness of Co layer is indicated in the figure ARTICLE IN PRESS N.N Phuoc et al / Journal of Magnetism and Magnetic Materials 298 (2006) 43–47 gradually reduces with the cycle of measurement causing the observed training-like effect Experiment The samples used in this work with the structure of Si(1 0)/CrMn (12 nm)/Co (x nm) were deposited at room temperature by the RF sputtering The CrMn layer was sputtered from a composite target constituting of Cr target with Mn chips placed on it The base pressure was about 10À6 mbar whereas the Ar pressure during deposition was 10À3 mbar The deposition was carried out without applying a magnetic field The composition of the CrMn films, identified by energy dispersive X-ray spectroscopy (EDS), is Cr45Mn55 The samples were then annealed in high 0.00030 0.002 0.001 M (emu) M (emu) 0.00015 0.00000 -0.00015 -1000 H (Oe) 1000 2000 -400 0.00030 0.002 0.00015 0.001 0.00000 T = 173 K -0.00015 -200 H (Oe) 200 400 0.000 -0.001 T = 173 K -0.002 -0.00030 -2000 T = 123 K -0.002 M (emu) M (emu) 0.000 -0.001 T = 123 K -0.00030 -2000 45 -1000 H (Oe) 1000 -400 2000 -200 H (Oe) 200 400 0.00030 0.002 0.001 M (emu) M (emu) 0.00015 0.00000 -0.00015 -0.001 T = 223 K T = 223 K -0.002 -0.00030 -2000 0.000 -1000 H (Oe) 1000 2000 -400 -200 H (Oe) 200 400 Fig Hysteresis loops of the CrMn(12 nm)/ Co(18 nm) (left panel) and CrMn(12 nm)/ Co(90 nm) (right panel) bilayers at various temperatures ARTICLE IN PRESS N.N Phuoc et al / Journal of Magnetism and Magnetic Materials 298 (2006) 43–47 vacuum oven (10À5 mbar) at the temperature of 300 1C for h They were subsequently cooled in the magnetic field of kOe to room temperature Magnetic properties of the annealed bilayers were characterized by vibrating sample magnetometer (VSM) in the temperature range from 123 K to room temperature 0.002 0.001 M (emu) 46 0.000 n=1 -0.001 n=5 Results and discussion The hysteresis loops of the annealed CrMn (12 nm)/Co (x nm) (x ¼ 18; 36, 54, 72, 90, 108 nm) bilayers, measured at 123 K were shown in Fig We have observed a very large exchange bias (H E % 600 Oe) It is noted that all the magnetization curves in Fig show double-loops and this effect is especially clear in the sample with tCo ¼ 72 and 90 nm Moreover, there is a strong correlation between the appearance of exchange bias and double-shifted loops as shown in Fig On the left panel of Fig are the hysteresis loops of the annealed CrMn (12 nm)/Co (18 nm) bilayer measured at different temperatures while on the right are those of the CrMn (12 nm)/Co (90 nm) bilayer measured at the same corresponding temperatures It is worth noting that the width of the kink in the curve with double-shifted loops decreases as the temperature increases in the same manner with the diminution of exchange bias and at the temperature of 223 K, both double-shifted loops and exchange bias disappear This suggests that the double-shifted loop may result from exchange bias Fig shows a representative of magnetic hysteresis loops of CrMn (12 nm)/Co (90 nm) film after some cycles of measurement It is remarkable that the right-hand loop becomes smaller and the left-hand loop becomes bigger with increasing n After nine cycles of hysteresis measurements, the right-hand loop disappears, making a singleshifted loop in the magnetization curve This effect seems to be similar to the training effect often observed in exchange bias system, in which exchange bias field decreases with the number of measurement [5] However, in our case, only the shape of the magnetization curve is changed with the cycle of measurement while the value of n=9 -0.002 -400 -200 200 400 H (Oe) Fig Representative of hysteresis loops of CrMn(12 nm)/Co (90 nm) film M–H loops for the cycles 1, 5, and are shown HFC Negative exchange bias FM AF (a) Positive exchange bias FM AF (b) Negative exchange bias Positive exchange bias FM AF (c) AF domain wall Fig Schematic diagram of the spin configurations of FM/ AF bilayers in three cases of HFC: (a) Small HFC making negative exchange bias, (b) Large HFC making positive exchange bias, (c) Intermediate HFC making a double-shifted loop ARTICLE IN PRESS N.N Phuoc et al / Journal of Magnetism and Magnetic Materials 298 (2006) 43–47 exchange bias field is nearly constant Moreover, after cycles of measurement, further measurement was carried out but the shift of the single loop was not changed any more This feature indicates that the physical origin of this traininglike effect is possibly different from that of the normal training effect The manifestation of double-shifted loop is presumably attributed to the overlap of positive and negative exchange bias Positive exchange bias is believed to be due to the fact that the interfacial interaction is antiferromagnetic and the origin of such antiferromagnetic coupling as proposed has been system specific [6] For small HFC, the magnetic hysteresis loop exhibits negative exchange bias with the AF spins at the interface aligned in the negative direction as in Fig 4(a) If HFC is large enough to align the AF surface magnetization along HFC as shown in Fig 4(b), thus overcoming the interface AF–FM interface magnetic interaction, the magnetization curve will exhibit a positive shift For intermediate HFC, the AF moments at the interface are partially aligned with the cooling field as sketched in Fig 4(c) causing the formation of two AF domains, one makes negative exchange bias and the other makes positive exchange bias The magnetization curve thus consists of two hysteresis loops, one shifted to negative and the other to positive, making a double-shifted loop The height of the positive loop depends on the portion of the AF moments aligned with the cooling field The fact that the right-hand loop, corresponding to positive exchange bias becomes smaller with the cycle of measurement suggests that this state is metastable The height of the positive loop decreases with the number of measurements as shown in Fig 3, implies that the area of the AF domain causing positive exchange bias diminishes correspondingly The training-like effect can therefore be described as the movement of the AF domain wall toward the positive-exchange-biased domain making this 47 domain shrink and finally be suppressed It is wellknown that the positive exchange bias is in a highenergy state so it is less stable than the negative exchange bias, which may explain why the training effect seems to affect only the right-hand side subloop Conclusion In summary, we have fabricated CrMn/Co bilayers with varied thickness of the Co layer and have observed the double-shifted loop in CrMn/Co bilayers with the thickness of Co layer in a narrow range This double-shift loop is assumed to be due to the coexistence of positive and negative exchange bias This assumption is consistent with a training-like effect observed for the first time in the sample exhibiting a doubleshifted loop Acknowledgements This work is supported by the State Programs on Fundamental Research of Vietnam under the Grant No 811604 References [1] W.H Meiklejohn, C.P Bean, Phys Rev 102 (1956) 141 [2] J Nogue´s, I.K Schuller, J Magn Magn Mater 192 (1999) 203 [3] J Nogue´s, D Lederman, T.J Moran, I.K Schuller, Phys Rev Lett 76 (1996) 4624 [4] I.V Roshchin, O Petracic, R Morales, Z.P Li, X Batlle, I.K Schuller, Los Alamos National Laboratory, Preprint Archive, Condensed Matter (2004), 1–14, arXiv:cond-mat/ 0411014 (http://xxx.lanl.gov/pdf/cond-mat/0411014) [5] K Zhang, T Zhao, H Fujuwara, J Appl Phys 89 (2001) 6910 [6] X Ke, M.S Rzchowski, L.J Belenky, C.B Eom, Appl Phys Lett 84 (2004) 5458 ... recently, Roshchin et al [4] have found that the state of coexistence of positive and negative exchange bias can be achieved by cooling the sample of FeF2 /Co in a properly chosen constant 0.00000... Positive exchange bias FM AF (b) Negative exchange bias Positive exchange bias FM AF (c) AF domain wall Fig Schematic diagram of the spin configurations of FM/ AF bilayers in three cases of HFC:... in CrMn/ Co bilayers with the thickness of Co layer in a narrow range This double-shift loop is assumed to be due to the coexistence of positive and negative exchange bias This assumption is consistent