The role of magnetic anisotropy in the Kondo effect

In the Kondo effect, a localized magnetic moment is screened by forming a correlated electron system with the surrounding conduction electrons of a non-magnetic host1. Spin S D 1=2 Kondo systems have been investigated extensively in theory and experiments, but magnetic atoms often have a larger spin2. Larger spins are subject to the influence of magnetocrystalline anisotropy, which describes the dependence of the magnetic moment’s energy on the orientation of the spin relative to its surrounding atomic environment3,4. Here we demonstrate the decisive role of magnetic anisotropy in the physics of Kondo screening. A scanning tunnelling microscope is used to simultaneously determine the magnitude of the spin, the magnetic anisotropy and the Kondo properties of individual magnetic atoms on a surface. We find that a Kondo resonance emerges for large-spin atoms only when the magnetic anisotropy creates degenerate ground-state levels that are connected by the spin flip of a screening electron. The magnetic anisotropy also determines how the Kondo resonance evolves in a magnetic field: the resonance peak splits at rates that are strongly direction dependent. These rates are well described by the energies of the underlying unscreened spin states.

Magnetism of Single Adatoms and Small Adsorbed Clusters Investigated by Means of Low-Temperature STM

Quentin Dubout

This thesis works reports on the magnetic properties of single transition metal atoms and hydrogenated compounds adsorbed on crystal surfaces. It is an experimental study, based on scanning tunneling microscopy (STM), initiated by the project of demonstrating the existence of magnetic remanence for single adsorbed atoms (adatoms). All the measurements have been performed with a scanning tunneling microscope operating at 0.4 K and equipped with 8.5 T superconducting coils. The investigated systems are cobalt atoms, pure and hydrogenated (cobalt hydrides), adsorbed on Pt(111) and graphene on Pt(111) surfaces. In the first case, we have immediately been confronted to the presence of cobalt hydrides on the surface. Three types of hydrides (CoH, CoH2, and CoH3) have been identified, quite different from the clean adatoms, and readily distinguished by their tunnel spectrum (STS). Twoofthesehydridesdisplaylow-energyvibrationalmodes(1−30meV),identifiedbytheir absence of response to a magnetic field, the effect of an isotopic substitution, and by comparison with known systems. The three hydrides can be deprotonated applying a sufficient voltage between tip and sample. CoH2 on Pt(111) presents a Kondo resonance, due to a half-integer spin (S = 3/2) and a peculiar magnetic anisotropy. This system corresponds to the first demonstration of the appearance of the Kondo effect upon hydrogen adsorption. The results of several attempts to observe remanence for single cobalt adatoms on Pt(111) with spin-polarized scanning tunneling microscopy (SP-STM) are also presented and discussed. The second system differs from the first one in the presence of a graphene layer on the Pt surface. The first experimental determination of the magnetic moment and magnetic anisotropy of adatoms on graphene is reported in this thesis. A magnetic moment of (2.2 ± 0.4) μB was measured by spin-excitation spectroscopy (SES), as well as a hard magnetization axis with an anisotropy energy of 8.1 ± 0.4 meV. This value, comparable to the record value of single cobalt atoms on Pt(111), has been mainly attributed to the strong hybridization between cobalt and graphene. Ab initio calculations confirmed our observations. Graphene thus represents an extremely promising substrate for future applications in nanoscale magnetism and spintronics. This system also exhibits three different cobalt hydrides in addition to the clean atoms. Their respective chemical composition, CoH, CoH2, and CoH3, was determined by direct comparison between high resolution STM images and simulated images obtained from ab initio calculated atomic structures. A controlled deprotonation method is presented. CoH3 is the only hydride to display spin excitations. Its magnetic moment is comparable to that of clean cobalt, it presents a hard magnetization axis as well, however, its magnetic anisotropy energy is sensibly lower, 1.7 ± 0.05 meV. The other two hydrides display vibrational excitations only. The effect of hydrogen on the magnetic properties of metal adatoms is therefore very strong in this system as well, and must be taken into account in future studies, hydrogen being a very common contaminant in ultra-high vacuum set-ups. Finally, a detailed study of the structure graphene mono- and bilayers on Ru(0001) was realized. Two different stackings, AB and AA, have been identified for the bilayers, the last one not observed to date for this system. The moiré structure of monolayer graphene and of this last type of bilayer has been determined: (11.57×11.57)R4.3◦ and (10.54×10.54)R4.7◦, respectively. Moreover, C-C bond distortions of more than 10%, observed in the STM images, have been studied in detail. A quantitative comparison with ab initio calculations allowed to identify their origin as the tilt of the graphene π-orbitals. These distortions are therefore purely artificial. This is an important result for future scanning tunneling microscopy studies of graphene.

EPFL2013

Magnetism of individual adatoms and of epitaxial monolayers

Anne Lehnert

This thesis reports on the magnetism of low coordinated Fe and Co atoms in different chemical and electronic environments. The objective of this research was to contribute to the fundamental understanding of the magnetic properties, such as magneto-crystalline anisotropy energy (MAE), orbital and spin magnetic moment, of surface supported transition metal (TM) atoms and one atomic layer thick epitaxial TM films. A detailed knowledge on the interplay of electronic hybridization and magnetic properties is mandatory to elaborate new materials with high perpendicular magnetic anisotropy and acceptable writing field for future use in magnetic recording devices. The magnetic properties of the different experimental systems were investigated by x-ray absorption spectroscopy, x-ray magnetic circular dichroism, and magneto-optical Kerr effect (MOKE) and they were correlated to the morphology obtained by scanning tunneling microscopy (STM). The first part focuses on the effect of electronic hybridization between isolated 3d TM atoms and different kinds of substrates. First, we considered 4d and 5d TM substrates known for their high magnetic polarizability due to the close onset of ferromagnetism. Deposition of single Fe and Co atoms on Rh(111), Pd(111), and Pt(111) produces an increasing MAE going from Rh to Pd and finally to Pt that we directly correlate to the increasing spin-orbit constant of the substrate interlinking the induced magnetic moment and the MAE. Further we investigated single Fe and Co atoms on two different insulating layers, namely Al2O3/Ni3Al(111) and Cu2N/Cu(001), providing an efficient decoupling of the TM 3d states from the conduction electrons of the substrates. In case of the alumina film the absorption spectra of Fe and Co show a pronounced multiplet structure. On Cu2N the multiplet features of the adatoms are less defined due to the lower thickness of the screening layer. On both insulating layers we find a common out-of-plane easy axis, highly unquenched orbital moments, and giant MAE values of several meV/atom which are due to the formation of strong anisotropic bonds with the O and N atoms, respectively. The second part is dedicated to single atom thick layers of Fe and Co deposited on Pt(111) and Rh(111). We find the MAE to be strongly enhanced compared with bulk and we establish a direct proportionality between the induced magnetic moment of the topmost substrate layer and the MAE. The study of bimetallic Fe1-xCox alloys on platinum reveal a material with high MAE (0.5 meV at the equiatomic composition), large saturation magnetization (3μB to 2μB going from Fe to Co), and perpendicular magnetization easy axis. These characteristics hint at new possibilities for the further miniaturization of the grain size in recording devices. In the last part the theoretical description of the MOKE for ultra-thin films is extended to the transverse geometry and improved analytic formulae for the polar and longitudinal geometry are presented. This finding was used to optimize the Kerr signal for a new MOKE setup which is fully integrated in an ultrahigh vacuum chamber combining in situ MOKE and STM. The setup performance was demonstrated for a continuous film of 0.9 monolayers (ML) Co/Rh(111) with in-plane easy axis and for a superlattice of nanometric double layer Co islands on Au(11,12,12) with out-of-plane easy axis. The detection limit is 0.5 ML for transverse and 0.1 ML for polar Kerr geometry corresponding for island superlattices with the density of Co/Au(11,12,12) to islands composed of about 50 atoms. Moreover, the setup holds two independent ways to measure the MAE which is either by thermally induced magnetization reversal or by applying a torque onto the magnetization by turning the magnetic field. Both ways yield similar results for 1.1 ML Co/Au(11,12,12).

EPFL2009

Propriétés magnétiques de nanostructures de cobalt adsorbées

Nicolas Weiss

This thesis reports results on magnetic properties of supported cobalt nanostructures. The nanostructures were grown on single crystal metal surface by Molecular Beam Epitaxy in an Ultra High Vacuum chamber. Our experimental setup is based on two in situ measurement techniques. The first is the Scanning Tunneling Microscope (STM) allowing to investigate the nanostructure morphology. The second, called Surface Magneto-Optical Kerr Effect (SMOKE), probes the magnetism of the nanostructures. By combining the data obtained with both techniques we were able to investigate the magnetism of 2D nanoparticles down to the atomic level. The first aim of the thesis consisted in disclosing the role played by the differently coordinated atoms in determining the Magnetic Energy Anisotropy in 2D nanostructures. To this purpose, we performed experiments on the Co/Pt(111) system, chosen as a model system since the Pt surface is well known to increase the out-of-plane uniaxial anisotropy as well as the magnetic moment of the Co islands. A preliminary work has been devoted to the growth of the Co islands as a function of the deposition and annealing temperatures. As a result, we learned to grow three different island's shapes : ramified, compact and double-layer islands. We then concentrated on the magnetism of these islands. Due to the nonlinear relationship between perimeter length and surface area, we were able to distinguish the different contributions to the anisotropy energy of surface and perimeter atoms. We found that the magnetic anisotropy in 2D Co nanostructures is predominantly due to the low-coordinated edge atoms, for which the anisotropy energy can be as large as 20 times the bulk value. This finding opens new possibilities to separately tune the anisotropy and moment of nanostructures. To exemplify this, and to illustrate once more the role of perimeter atoms, we produced Co island with a non-magnetic Pt core. As expected, these bimetallic islands have identical anisotropy and lower magnetic moment to their equally shaped pure Co counterpart. The second aim we focused on was the investigation of the magnetic properties of ultra-high density arrays of nanostructures created by self-assembly. We chose Co/Au(788) as a model system since the nanoscopic Co dots were found to have a very narrow size distribution and to be placed onto a lattice which is phase coherent on a macroscopic length scale. Moreover, the array has a density of 26 Tdot/in2 which allows the investigation of inter-particle interaction in ultra high density arrays. We showed that the narrow size distribution of the dots corresponds to an unprecedented narrow Magnetic Anisotropy Energy distribution with 35% width at half of the maximum. Moreover, we demonstrated the weak magnetic coupling between the Co dots despite their high density. In parallel to these fundamental researches, we developed a new experimental setup that will replace the current one. We designed a UHV chamber equipped with a homemade variable temperature STM combined with a SMOKE experiment. A quadrupole, focused on the sample, generates a 0.35 T magnetic field. This field is one order of magnitude larger than the present one and can be uniformly rotated in the plane of the quadrupole. Furthermore, the design of the new chamber allows STM imaging in this magnetic field. In this report, we present the adopted solutions as well as the first tests of the components (STM, SMOKE, cryostat).

EPFL2004