Thin-film structures presenting perpendicular magnetic anisotropy (PMA) are widely studied for applications in high-density magnetic recording. PMA, with the film magnetisation pointing spontaneously along the perpendicular to the film, is an exceptional behaviour, as the shape of the thin film strongly favours magnetisation parallel to the film. Pt-Ni multilayers present PMA at room temperature [1-3], however PMA appears when the Pt layers are very thin (~ 2 monoatomic layers (ML) thick). This is in contrast with Co-Pt multilayers where the Ni layers are very thin. PMA in Pt-Ni systems has been attributed to a bulk-like magnetoelastic effect in the Ni layers [3], but it is still unclear why very thin Pt layers are necessary.

 

Fig. 93: (a) Schematics of the stacking sequence of the FCC (111) planes in a Pt-Ni multilayer. The [1-11] direction does a "zig-zag" when going from one layer to the next. This sequence gives a hexagonal symmetry to the multilayer, which, combined with the full magnetisation of very thin Pt layers, leads to perpendicular magnetic anisotropy. (b) Surface diffraction data showing the stacking conservation for Ni on Pt(111) and the stacking reversal for Pt on Ni (111). The comparison of the substrate and film diffraction rods enables are comparison of the stackings (ABC or ACB) of the substrate and the film, since the Bragg peaks for the two stacking appear at different L's on this rod.

 

Here we propose a new structural explanation for the PMA, that also explains the need for very thin Pt layers. This is the stacking reversal. The idea is that, upon growing Pt-Ni multilayers, Pt and Ni, which have a FCC structure, both grow with the [111] axis along the growth direction. They can a priori grow either with an ABCABC... stacking of the (111) planes, or a ACBACB... stacking. Usually, films tend to grow with the same stacking sequence as the substrate. We show that this is the case for Ni on Pt(111) (normal growth), but not for Pt on Ni(111), which grows with reversed stacking. As a consequence, Ni-Pt multilayers, will exhibit alternate packing sequences of the Ni layers as normal/reverse/normal/reverse (Figure 93a).

Our study was performed on the ID03 beamline, with only single interfaces, Pt films on a Ni(111) single crystal, and Ni films on a Pt(111) single crystal. We investigated the structure of the films with standard surface diffraction, and showed the stacking conservation for Ni on Pt, and the stacking reversal for Pt on Ni (Figure 93b). With resonant magnetic surface X-ray diffraction at the Pt LIII edge, we determined the profile of induced magnetic moment in the Pt near the interface (the "positions" of the magnetic moments). We showed that the Pt moment is concentrated in the Pt plane in contact with the Ni, for both a 8 ML Ni film on Pt (Figure 94) and a 4 ML Pt film on Ni.

 

Fig. 94: Standard truncation rods of the Pt substrate (filled circles) and corresponding magnetic rods (open circles), for a 8 layer Ni film on Pt(111). The magnetic rods are obtained by measuring, at each L, the diffracted intensity for two directions of the applied magnetic field (pointing up, or pointing down), and forming the asymmetry ratio R(HKL)= (I-I)/(I+I). The atomic positions are first determined using the standard diffraction rods, then the atomic model is re-injected into the calculation of the magnetic rods, and from a fit we deduce the magnetic moments of the different Pt planes near the interface. The calculated magnetic rod (solid line) was obtained assuming that only the last Pt plane is magnetic, with a moment of 0.06 mB/atom at 300K.

 

We can explain how stacking reversal induces PMA with a symmetry argument: if one forgets about the Pt layers, and considers only the stacking of the Ni (111) planes, then the Ni-Pt multilayer is analogous to a thick Ni film with a hexagonal structure, with the c axis perpendicular to the layers. This is not HCP Ni but a hexagonal polytype with a period of two Ni layers. By analogy with the case of Co, whose HCP phase (easy axis along c) has a much larger magnetic anisotropy than the FCC phase (easy axes along <111>), this hexagonal Ni is expected to have a much larger magnetocrystalline anisotropy than FCC Ni, with an easy axis along c. This should favor PMA.

We think that very thin Pt layers are necessary for PMA because in order for the multilayer to act as a high magnetic anisotropy hexagonal material and not as quasi-isotropic "isolated" FCC Ni layers, the adjacent Ni layers need to exchange-coupled through the Pt layers. For this the Pt layers need to be fully magnetic. This is only true for Pt layers with a thickness of two atomic planes or less: we show that, at the Pt-Ni interface, the magnetisation of the Pt atoms is essentially concentrated in the Pt plane in contact with the Ni, the next Pt plane has a magnetisation at least three-times smaller.

References
[1] M. Angelakaris, P. Poulopoulos, N. Vouroutzis, M. Nyvlt, V. Prosser, S. Visnovsky, R. Krishnan, N.K. Flevaris, J. Appl. Phys. 82, 5640 (1997).
[2] F. Wilhelm, Ph.D. thesis, dissertation.de, Berlin (2000).
[3] S.C. Shin, G. Srinivas, Y.S. Kim, M.G. Kim. Appl. Phys. Lett. 73, 393 (1998).

Principal Publication and Authors
O. Robach (a), C. Quirós (a), H. Isérn (a), P. Steadman (a,b), K.F. Peters (a,c), S. Ferrer (a), Phys. Rev. B 67 (2003) 220405(R).
(a) ESRF
(b) University of Leeds (UK)
(c) Hewlett Packard, Corvallis (USA)