# Dresden 2017 – wissenschaftliches Programm

## Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

# O: Fachverband Oberflächenphysik

## O 18: Semiconductor Substrates: Structure, Epitaxy and Growth

### O 18.1: Vortrag

### Montag, 20. März 2017, 16:00–16:15, WIL C307

**Deconstructing silicon: can surface reconstructions emerge from global geometry screening?** — •Chiara Panosetti^{1}, Maximilian Bauer^{1}, Matt Probert^{2}, and Karsten Reuter^{1} — ^{1}Technische Universität München, Germany — ^{2}University of York, GB

The *Si*(111)−(7 × 7) reconstruction is arguably the most complicated surface superstructure known to date. The details of its currently accepted geometry, described by the so-called Dimer-Adatom-Stacking fault (DAS) model, are the result of a 25-year long joint effort of experiments and theory. Attempting to solve such a complicated problem—with a known solution—by means of unbiased global structure optimization is therefore a suitable testbed for the performance of corresponding computational algorithms. We here specifically assess the performance of Genetic Algorithms (GA) and the Basin Hopping (BH) approach, where the generation of trial structures for canonical and grand canonical screening is applied to the surface layer in periodic boundary conditions and where different strategies are pursued to enhance the production of chemically sensible candidate geometries. Results suggests that robustness, success and speed of convergence of the employed approaches are strongly influenced by how much the trial moves tend to preserve favourable bonding patterns once they appear.