We report on the magnetic properties of two-dimensional Co nanoparticles arranged in macroscopically phase-coherent superlattices created by self-assembly on Au(788). Our particles have a density of 26 Tera=in 2 (1 Tera 10 12 ), are monodomain, and have uniaxial out-of-plane anisotropy. The distribution of the magnetic anisotropy energies has a half width at half maximum of 17%, a factor of 2 more narrow than the best results reported for superlattices of three-dimensional nanoparticles. Our data show the absence of magnetic interactions between the particles. Co=Au788 thus constitutes an ideal model system to explore the ultimate density limit of magnetic recording. DOI: 10.1103/PhysRevLett.95.157204 PACS numbers: 75.75.+a, 75.30.Gw, 81.16.Dn The bit density on magnetic hard disks has been increasing at a constant pace for many years [1]. Besides the technological challenges, we face today the question of where downscaling ends from fundamental physics. This question can be addressed by studying periodic lattices of ferromagnetic monodomain particles, where each particle stores one magnetic bit. The high density requires out-ofplane magnetic anisotropy to minimize dipolar interactions among adjacent particles. Further, to optimize the signalto-noise ratio in read or write processes, the magnetic properties have to be uniform; i.e., the particles have to be uniaxial and the distributions of moments M, and magnetic anisotropy energies (MAEs), K, have to be narrow.Chemical synthesis of self-assembled colloid particles has led to excellent size distributions with a half width at half maximum (HWHM) of 7% in diameter and 21% in volume, respectively [1,2]. However, these are accompanied by systematically much wider K distributions [3,4] partly caused by the random orientation of the particle's easy axes [4,5] causing strong dipolar interactions. Colloid particles have obvious practical advantages, one of them being the high blocking temperatures [6]. However, for the outlined reasons, the ultimate density limit could not be explored with such systems so far. Alternatively, massselected magnetic 3D clusters [7] are monodisperse and may be soft-landed onto surfaces [8], but cannot be arranged into ordered arrays of equidistant magnetic units.Self-assembly during atomic beam epitaxy on periodic strain relief patterns on single crystal surfaces has been shown to provide well ordered superlattices of 2D islands with narrow size distributions [9]. While such lattices are generally not in phase from terrace to terrace, a combination of epitaxial strain relief patterns and vicinal surfaces has been shown to lead to macroscopically phase-coherent lattices [10]. Former studies of the magnetic properties of metal islands on single crystal surfaces revealed that the spin-orbit interaction with the substrate may induce a strong enough magnetocrystalline anisotropy to yield a common out-of-plane easy magnetization axis for all islands [11][12][13][14]. Therefore we have today the tools to fabricate the required lattices of uni...

This paper reports on Au(111) vicinal surfaces, either regularly stepped surfaces, reconstructed or not, or periodically faceted surfaces, which are well suited to be used as templates for organized growth of clusters. Angles of misorientation with respect to the (111) plane lie between 1 • and 12 • and two opposite azimuths are considered: (i) [211], that leads to steps with {100} microfacets, and (ii) [211], that leads to steps with {111} microfacets. The behaviour of the Au(111) reconstruction in the vicinity of steps depends drastically on the step microstructure, and this is a key point for understanding the various periodic morphologies existing on Au(111) vicinal surfaces. The interaction between the reconstruction and the close-packed steps of the Au(111) surface is interpreted in terms of the relative stability of both types of step. Self-organized morphologies between 10 and 100 nm are interpreted within the framework of elastic theory and by pointing out the crucial role played by the atomic boundary energy term. The microscopic origin of faceting is discussed, proposing two different models depending on each azimuth. Then, we illustrate the use of Au(111) vicinal surfaces as templates for growing long range ordered nanostructures. Examples are given in the case of cobalt growth.

A novel approach based on strain-relief vicinal patterned substrates is demonstrated for two-dimensional (2D) self-ordered growth. Long-range-ordered cobalt nanodots growth is achieved using a Au(788) vicinal surface which is spontaneously patterned in two dimensions at a nanometer scale with a macroscopic coherence length. Performing the cobalt growth on a substrate cooled down to 130 K allows a high degree of nanostructure uniformity. The atomic processes responsible for such a behavior are discussed. Such a high quality of both long-range-and local-ordered cobalt growth opens up the possibility of making measurements of physical properties of such nanostructures by macroscopic integration techniques.