First-principles modeling of molecular switches at surfaces

Abstract

The scientific vision of nanotechnology is the atomically precise fabrication and manipulation of mechanical and electronic components. A promising route to such control on the molecular scale, is to construct components from controllable molecules. Of these, molecules with properties bi-stably and reversibly modifiable by external stimuli, so-called molecular switches, are the simplest case with obvious applications. The azobenzene molecule qualifies in this class by undergoing reversible photo-isomerization between its cis and trans conformers with particularly high yield and stability, rendering it an archetype of molecular switches research. Ostensibly, direct interaction with single or few switch units requires localization and ordering of switches, e.g. adsorbed at a solid surface. However, so far, surface adsorption of azobenzene without substantial to total loss of switching function has not been achieved. The use of ligands decoupling switch moieties from the substrate, while a promising possible solution to this problem, has a comparably detrimental influence in a number of cases. This thesis investigates azobenzene adsorption in the two model cases of complete and no switch substrate decoupling, using theoretical surface science techniques. Aiming at quantitatively predictive modeling, a wide range of first-principles and ab initio simulation methods is employed. Since in particular interactions of organic molecules with metal surfaces pose a tremendous challenge to such methods, the second main theme is methodological in nature, notably addressing the problem of simultaneous treatment of a metallic substrate bandstructure, and weak van der Waals (vdW) substrate adsorbate interactions. In combination with various experimental X-ray and UV/Vis spectroscopy techniques, azobenzene-functionalized self-assembled monolayers (SAMs) are studied as an example of complete switch surface decoupling. The results identify excitonic coupling between switch units as a main cause of switch yield loss in this system. Implying that such loss is intrinsic to azobenzene above a critical component density, this finding – in addition to previously discussed steric limitations on switch motion – is crucial to future design of surface-decoupled switch arrays. Direct switch adsorption at close-packed coinage metal (Cu, Ag, Au) surfaces is modeled explicitly accounting for the substrate electronic structure. The current work-horse simulation method for such a problem – density-functional theory (DFT) with (semi-) local exchange-correlation functionals – is shown to yield qualitatively incorrect results, primarily due to its deficient description of vdW interactions. Currently beyond the capabilities of accurate ab initio techniques, the problem is revisited using DFT with semi-empirical correction potentials (DFT-D), resulting in a more plausible and consistent bonding picture at all three substrates. State-of-the-art X-ray spectroscopy experiments find the geometry prediction of the most sophisticated correc-