Coordination dependent magnetic properties of 3d and 4d metal nano-structures

Abstract

In this thesis, the magnetic properties of self-assembled 3d and 4d metal nano-structures supported on surfaces have been investigated. The atomic coordination within the nano-structures was found to profoundly affect important quantities such as the magnetic moment and the magnetic anisotropy. The use of thin, atomically flat insulating Xe spacers of 1-15 monolayers (ML) thickness allowed for a study of coordination effects in the two limits of strong and weak coupling with an underlying metal substrate. The systems were characterized by surface-sensitive methods, based on synchrotron radiation (X-ray magnetic circular dichroism, and X-ray scattering/diffraction) and variable temperature scanning tunneling microscopy (VT-STM). The VT-STM was developed and implemented during this PhD work. First, the magnetism of Rh nano-structures on a Xe buffer layer has been investigated. Rh is non-magnetic in bulk but shows a finite magnetic moment upon reducing cluster sizes to below 100 atoms. Within this work a small, non-zero magnetic moment was found for Rh nano-structures situated on Xe. The effect of intra-cluster Rh-Rh coordination was observed to affect both the spin and orbital part of the magnetic moment, leading to strongly oscillating values at smallest cluster sizes. Further, the analysis of the spectroscopic data suggests an interpretation for the absence of magnetism in directly deposited Rh on Ag(100) that is based on the formation of a kinetically promoted Ag-Rh alloy. Second, the buffer layer assisted growth (BLAG) was studied for sub-monolayer Co nano-clusters on Ag(111) and Pt(111) surfaces. The observation of the cluster formation process in the very early stages of BLAG revealed the paramount importance of the substrate in determining both magnetism and structural properties of the nano-clusters. On Ag(111), a weakly interacting substrate, the clusters form on the buffer layer independently from the metal substrate and show no magnetic anisotropy at this stage. As soon as the Xe is desorbed by sample annealing an in-plane anisotropy forms, as a consequence of the contact with the substrate. X-ray scattering and diffraction data support this interpretation and also show that in the limit of a single monolayer of Xe on Ag(111) the BLAG is a 'simple' atomic diffusion process, with a very high mobility of Co atoms on Xe. On a thick Xe buffer layer instead, due to a lower Xe-Xe binding energy and to the higher surface energy of Co compared to Xe, the deposition of Co provokes a re-arrangement of the Xe atoms. On the other hand, on Pt(111) the BLAG process fails to ensure a cluster formation process ontop the buffer layer and independent of the metal substrate. Here in fact, electric dipolar interactions occurring between Co atoms and the substrate through the Xe layer, are strong enough to destroy the Xe ML order and bring the Co atoms in direct contact with the Pt(111) before Xe atoms are thermally desorbed. This complex process becomes evident from VT-STM investigations and by the occurrence of perpendicular magnetic anisotropy right after Co deposition on the Xe ML/Pt(111). In a detailed discussion it is shown that magnetic properties like magnetic anisotropy and orbital/spin moments are strongly entangled with their morphology. Both morphology and magnetism are determined by the interaction with the environment. This opens the way to more complex systems, where the interaction with the medium is tuned such as to gain nano-structures with a pre-defined structure and function. Third, the knowledge about the cluster-substrate interactions during BLAG was exploited to build highly ordered arrays of Co nano-structures on a patterned template substrate. In this case the hexagonal boron nitride (h-BN) nanomesh on Rh(111) was used. These systems have been employed to study the effect of hybridization of the Co d band with capping layers such as Pt, Au, Al2O3 and MnPt on the magnetic moment of Co. It was found that in all these cases Co clusters have no remanence, due to the small size and weak coupling with the h-BN atoms. However, it could be shown that capping the clusters strongly influence the clusters magnetization, in a non-trivial way.