No Arabic abstract
Recent photoemission experiments on the Si(553)-Au reconstruction show a one-dimensional band with a peculiar ~1/4 filling. This band could provide an opportunity for observing large spin-charge separation if electron-electron interactions could be increased. To this end, it is necessary to understand in detail the origin of this surface band. A first step is the determination of the structure of the reconstruction. We present here a study of several structural models using first-principles density functional calculations. Our models are based on a plausible analogy with the similar and better known Si(557)-Au surface, and compared against the sole structure proposed to date for the Si(553)-Au system [Crain JN et al., 2004 Phys. Rev. B 69 125401 ]. Results for the energetics and the band structures are given. Lines for the future investigation are also sketched.
Using x-ray diffraction Ghose et al. [Surf. Sci. {bf 581} (2005) 199] have recently produced a structural model for the quantum-wire surface Si(553)-Au. This model presents two parallel gold wires located at the step edge. Thus, the structure and the gold coverage are quite different from previous proposals. We present here an ab initio study using density functional theory of the stability, electronic band structure and scanning tunneling microscopy images of this model.
When gold is deposited on Si(553), the surface self-assembles to form a periodic array of steps with nearly perfect structural order. In scanning tunneling microscopy these steps resemble quasi-one-dimensional atomic chains. At temperatures below ~50 K the chains develop tripled periodicity. We recently predicted, on the basis of density-functional theory calculations at T=0, that this tripled periodicity arises from the complete polarization of the electron spin on every third silicon atom along the step; in the ground state these linear chains of silicon spins are antiferromagnetically ordered. Here we explore, using ab-initio molecular dynamics and kinetic Monte Carlo simulations, the behavior of silicon spin chains on Si(553)-Au at finite temperature. Thermodynamic phase transitions at T>0 in one-dimensional systems are prohibited by the Mermin-Wagner theorem. Nevertheless we find that a surprisingly sharp onset occurs upon cooling---at about 30 K for perfect surfaces and at higher temperature for surfaces with defects---to a well-ordered phase with tripled periodicity, in good agreement with experiment.
We present here a comprehensive search for the structure of the Si(553)-Au reconstruction. More than two hundred different trial structures have been studied using first-principles density-functional calculations with the SIESTA code. An iterative procedure, with a step-by-step increase of the accuracy and computational cost of the calculations, was used to allow for the study of this large number of configurations. We have considered reconstructions restricted to the topmost bilayer and studied two types: i) flat surface-bilayer models, where atoms at the topmost bilayer present different coordinations and registries with the underlying bulk, and ii) nine different models based on the substitution of a silicon atom by a gold atom in different positions of a $pi$-bonded chain reconstruction of the Si(553) surface. We have developed a compact notation that allows us to label and identify all these structures. This is very useful for the automatic generation of trial geometries and counting the number of inequivalent structures, i.e., structures having different bonding topologies. The most stable models are those that exhibit a honeycomb-chain structure at the step edge. One of our models (model f2) reproduces the main features of the room temperature photoemission and scanning-tunneling microscopy data. Thus we conclude that f2 structure is a good candidate for the high temperature structure of the Si(553)-Au surface.
We present a systematic study of the atomic and electronic structure of the Si(111)-(5x2)-Au reconstruction using first-principles electronic structure calculations based on the density functional theory. We analyze the structural models proposed by Marks and Plass [Phys. Rev. Lett.75, 2172 (1995)], those proposed recently by Erwin [Phys. Rev. Lett.91, 206101 (2003)], and a completely new structure that was found during our structural optimizations. We study in detail the energetics and the structural and electronic properties of the different models. For the two most stable models, we also calculate the change in the surface energy as a function of the content of silicon adatoms for a realistic range of concentrations. Our new model is the energetically most favorable in the range of low adatom concentrations, while Erwins 5x2 model becomes favorable for larger adatom concentrations. The crossing between the surface energies of both structures is found close to 1/2 adatoms per 5x2 unit cell, i.e. near the maximum adatom coverage observed in the experiments. Both models, the new structure and Erwins 5x2 model, seem to provide a good description of many of the available experimental data, particularly of the angle-resolved photoemission measurements.
Stabilization of the Si(553) surface by Au adsorption results in two different atomically defined chain types, one of Au atoms and one of Si. At low temperature these chains develop two- and threefold periodicity, respectively, previously attributed to Peierls instabilities. Here we report evidence from scanning tunneling microscopy that rules out this interpretation. The x3 superstructure of the Si chains vanishes for low tunneling bias, i.e., close the Fermi level. In addition, the Au chains remain metallic despite their period doubling. Both observations are inconsistent with a Peierls mechanism. On the contrary, our results are in excellent, detailed agreement with the Si(553)-Au ground state predicted by density-functional theory, where the x2 periodicity of the Au chain is an inherent structural feature and every third Si atom is spin-polarized.