Magnetic data storage technology is driven by the quest to decrease the length scale of the magnetic bit cell. Since the typical layer thicknesses involved in devices has decreased to only a few nm, maintaining the stability of the magnetic state over time has become a key issue. Indeed, it is well known that a ferromagnetic layer enters a super paramagnetic state (i.e. the magnetic state is continuously changing) once its thickness decreases below a critical value. This poses a limit for the size of the bit cell that can be used in conventional magnetic recording.

Therefore, it is of particular interest to explore any possible mechanism that leads to stabilisation of magnetic order beyond the superparamagnetic limit. Of special interest in this field are oxide antiferromagnets, where long range magnetic correlations are maintained through nearest neighbour interactions that couple to adjacent ferromagnetic layers to form an exchange-bias system.

Following the pioneering work of Beach et al. who revealed intriguing magnetic properties of ultrathin native iron oxide layers [1], we studied the chemical and magnetic properties of oxidised iron layers in situ using nuclear resonant scattering (NRS) at beamline ID18. We used the isotope sensitivity of NRS to isolate the signal originating from the oxide layer. This proved to be of utmost importance to discriminate against the strong magnetic signal of the neighbouring Fe metal, which normally prevents the detection of the oxide’s magnetic state using conventional techniques. In an NRS experiment, the time delayed de-excitation of the 57Fe nucleus is recorded after excitation by pulsed synchrotron radiation. The magnetic state of the 57Fe atoms is encoded in the shape of the measured temporal beat patterns. For instance, a non-magnetic material would show only a simple exponential decay of the signal whereas a magnetic one shows a beat frequency which is a signature of its particular magnetic configuration.

Fig. 13: Evolution of the magnetic state of a thin FeO layer during growth. The layer is being stabilised magnetically by the surrounding ferromagnet. The corresponding NRS spectra and their fits are shown at the bottom.

This property was used to monitor the chemical and magnetic state of the growing oxide layer. The complete process is summarised in Figure 13. We start with a 2 nm 56Fe layer on top of which a 0.6 nm 57Fe layer (the NRS active isotope) is deposited. This layer is subsequently oxidised by controlled admission of oxygen into the vacuum chamber. A sudden change from pure magnetic Fe to a chemically disordered, predominantly non-magnetic oxide is observed. Remarkably, the deposition of one atomic layer of Fe on top leads to the formation of a chemically pure FeO layer [2], which is non-magnetic. This is not surprising as the bulk Néel temperature TN of FeO is found to be around 200 K. However, with further deposition of Fe, a sharp magnetic transition is observed in the oxide, which means that TN has effectively risen above room temperature. We expect that this magnetic transition coincides with the appearance of long range ferromagnetism in the top Fe layer, which is only possible once a critical thickness is reached. Ex situ temperature dependent measurements allowed the TN to be estimated as being as high as 800 K, very close to the Curie temperature TC of the neighbouring Fe layer. This similarity is not a coincidence, as the proximity of very thin ferromagnetic layers can lead to a convergence of TN towards the TC of the ferromagnetic material.

This study unravels how the magnetic state of ultrathin antiferromagnetic layers can be stabilised by an adjacent ferromagnet such that its effective Néel temperature can be increased by several hundreds of Kelvin. In the case of FeO, it leads to antiferromagnetic order at room temperature, an effect which proved to be very effective to mediate strong interlayer coupling in Fe/FeO heterostructures.



[1] G.S.D. Beach, F.T. Parker, D.J. Smith, P.A. Crozier and A.E. Berkowitz, Phys. Rev. Lett. 91, 267201 (2003).
[2] S. Couet, K. Schlage, K. Saksl and R. Röhlsberger, Phys. Rev. Lett. 101, 056101 (2008).


Principal publication and authors

S. Couet (a,b), K. Schlage (a), R. Rüffer (c), S. Stankov (c), Th. Diederich (a), B. Laenens (b) and R. Röhlsberger (a), Phys. Rev. Lett. 103, 097201 (2009)
(a) HASYLAB at DESY, Hamburg (Germany)
(b) Instituut voor Kern- en Stralingsfysica, K.U. Leuven (Belgium)
(c) ESRF