The constant demand for higher storage density in the magnetic device industry explains the recent and dramatic increase of interest in nanostructures. In order to produce nanoparticles with high magnetisation and stable properties the basic physics of these particles needs to be understood. As a model system, we have chosen the epitaxial growth of iron on a Cu(111) substrate. Depositing iron on a copper single crystal allows iron to be stabilised at room temperature in its fcc phase γ-Fe). For bulk Fe this phase only appears at temperatures (>1186 K) above the Curie temperature. Moreover, the magnetic structure of the γ-Fe strongly depends on its lattice constant. In general theory predicts for γ-Fe an increasing magnetic moment, as well as ferromagnetic coupling, with increasing lattice parameter [1]. Since the Cu substrate has a lattice parameter slightly bigger than γ-Fe at equilibrium, we can expect γ-Fe to have a higher moment for pseudomorphic growth.

To obtain nanostructures we take advantage of the step decoration effect of Fe grown on vicinal Cu(111). This surface is characterised by monoatomic steps and regular 10 nm wide terraces. As Fe atoms are deposited they tend to nucleate along the steps thus forming 1-D Fe stripes [2]. Going from a few percent of a monolayer (ML) to 4 ML, we cover the whole thickness range of interest for the γ-Fe. In particular below the 0.8 ML thickness range, the Fe forms essentially elongated clusters of low extension (less than 3000 atoms per cluster) and from 0.8 ML to ~1.5 ML the Fe forms 1-D stripes with increasing width, until 2-dimensional percolation occurs. Thus we expect a dependence of the magnetic properties both on size (growth of nanoparticles) and on structure γ-Fe magnetic properties to small structural changes).

The magnetic characterisation was done at the beamline ID12B using X-ray magnetic circular dichroism (XMCD) at the Fe L2,3 absorption edges. An example for 0.13 ML of Fe/Cu(111) can be seen in Figure 68. The dichroic signal, i.e. the difference between two absorption spectra taken with opposite circular polarisation, allows a quantitative determination of the magnetic moment using magneto-optic sum rules.

In Figure 69 we plot the magnetic spin (mspin) and orbital (morb) moments per hole in the 3d band, as deduced from our XMCD measurements. Several observations can be made: The increase of the magnetic moment above ~2 ML is due to the start of the fcc to bcc phase transition. The bcc phase with larger magnetic moments dominates at 4 ML where we obtain the values for bulk bcc Fe. The clear increase of the orbital moment at low coverage is due to size effects, as it is well known that a more atomic-like structure, like at a surface or in a cluster, leads to an enhancement of the orbital moment. However, the most striking result is the observation of two low-spin phases, LS1 and LS2 (Figure 69), above and below 0.8 ML thickness that corresponds to the coalescence of the clusters into 1-D stripes. The presence of these two magnetic phases is tentatively ascribed to a structural relaxation: the Fe clusters, purely pseudomorphic at low coverage (aFe = aCu = 3.61 Å), undergo a strain relaxation to their stable value (aFe = 3.59 Å) once they coalesce and form a larger structure. This decrease in the volume could explain the decrease in mspin for the LS2 phase.

In conclusion, this type of study can lead to a basic understanding of the effect of size and structural changes caused by reducing the size of magnetic nanoparticles.

[1] V. L. Moruzzi, P.M. Marcus, J. Kübler, Phys. Rev. B, 39, 6957 (1989).
[2] J. Shen, M. Klaua, P. Ohresser, H. Jenniches, J. Barthel, Ch.V. Mohan, J. Kirschner, Phys. Rev. B, 56, 11134 (1997).

P. Ohresser (a), G. Ghiringhelli (a), O. Tjernberg (a), M. Finazzi (b), N.B. Brookes (a).

(a) ESRF
(b) TASC-INFM, Elettra Synchrotron Light Source, Trieste (Italy)