Fig. 1: (a) TEM view of 5 nm Co clusters deposited on Si; (b) HRTEM picture of a 4 nm Co cluster, showing the distinctive pattern of an icosahedral structure along a threefold axis, as shown in (c).
Nanometric clusters are one of the building blocks for many emerging technologies in the fields of spintronics, health sciences or energy. All these applications rest on specific structural, magnetic, electronic, chemical or physico-chemical properties brought about by small size or surface effects, many of them still poorly known or difficult to implement. NM has developed a powerful and versatile tool for the growth of cluster-based materials and devices. The chief advantage of this apparatus is its ability to grow size-controlled clusters of virtually any composition, coupled with the ability to insert them in sputtered thin films in order to either obtain new functional material or to perform state of the art measurements.
The magnetic anisotropy of a magnetic material is, often more than its intrinsic magnetization, the key parameter governing its usefulness for practical application. However, whereas in clusters the magnetization is most of the time close to that of the bulk, the magnetic anisotropy differs significantly, being very often much smaller than what would have been required for room temperature applications.
During the last few years we have carried out an extensive study of the anisotropy of clusters, with a strong emphasis on cobalt. One fundamental drive for this was the observation that the gas condensation process used for clusters growth leads to a peculiar multiply-twinned structure with an overall icosahedral shape, as shown in figure 1. These particles are thus a perfect case study for the understanding of size and structure effects.
This structure can be viewed as an assemblage of 20 distorted twinned fcc crystallites, with a multiplicity of equivalent symmetry axes: One question that arises is the consequence of such a structure on the anisotropy taking into account the fact that high anisotropy goes hand in hand with low crystal symmetry. By comparing magnetic measurements on assemblies of clusters with micromagnetic computation based on realistic atomic structure – resulting from a tight-binding calculation – it was shown that large and rapid variation in the anisotropy results from the fact that for clusters with a closed outer atomic shell the anisotropy is down to very low values, whereas for intermediate filling of the surface the effective anisotropy is almost always below that of the bulk.
Fig. 2: Polar representation of the switching field for one single Co/CoO core – shell cluster. Maximum measured field is 0.34 T. (a) MicroSQUID measurement after field-cooling along 0°; (b) Calculated curve taking into account the contributions from one F and two AF sub-lattices. The dashed line is the ferromagnetic anisotropy easy direction; the arrows indicate the directions for the AF sub-lattices; (c) Schematic view of the 3D switching curve surface for a uniaxial ferromagnet, cut by the microSQUID’s measurement plane.
Even though low magnetocrystalline anisotropy might hinder the use such particles in real-life devices, much higher anisotropy can be expected from the surface. This is especially true with Co/CoO clusters where the antiferromagnetic (AF) CoO shell, with anisotropy more than ten times that of cobalt, is exchange-coupled with a ferromagnetic (F) Co core. We succeeded in growing core-shell clusters in a very controlled manner, achieving fine tuning of the core and shell properties: Magnetic characterization of such particles reveals, for instance, that the AF magnetic shell of isolated clusters is too defective to provide an effective exchange coupling. This last observation relies among other things on measurement on single clusters using the microSQUID technique. For this we develop devices where the core-shell clusters or clusters on AF are embedded in micron-size niobium Josephson junctions. By measuring the switching field of individual clusters it was possible to determine the magnetic configuration of the F core and the two AF sublattices (Fig. 2), the latter being far less anisotropic than bulk CoO. It is worth mentioning that the sensitivity for this measurement was 109 times better than that of a conventional SQUID.
Following this, due also to recent improvements in the microSQUID technique allowing for the measurement of the magnetization curve of individual clusters, we began the study of clusters exchange-coupled with a single crystalline AF substrate in order to fully characterize the interplay between the structural properties and the magnetic behaviour. This constitutes a pre-requisite for the control of the exchange coupling induced anisotropy.
 R. Morel et al., Phys. Rev. Lett., 97 127203 (2006).
 C. Portemont et al., Phys. Rev. B, 78 144415 (2008).
Last update : 10/17 2013 (480)