Mechanisms of Local and Nonlocal Atomic Manipulation with a Scanning Tunnelling MIcroscope

Abstract

This thesis presents a detailed study of the physical processes underpinning manipulation of aromatic molecules with the scanning tunnelling microscope (STM) on Si(111)-7x7.We distinguish between two modes of manipulation: local and nonlocal.Nonlocal manipulation is when an electron injected from the STM tip induces amolecule some tens of nanometres distant from the original injection site to react,in this case desorb. We split this process into three steps: (i) electron injection, (ii)surface transport, and (iii) molecular manipulation. The first set of results presented in this thesis aim to gain comprehensive understanding of the transport process, step (ii). By drawing a comparison with laser two-photon photoemission experiments, we are able to show that the injected electrons are transported across the surface via a state localised at the silicon adatom backbonds. We also show that the temperature dependence of the length-scale of the nonlocal process obeys Einstein's electrical mobility equation, T^{0.5}. This, combined with a good t to the radial distribution data allows us to conclude that the observed nonlocal effect is the aftermath of diffusive hot electron transport across the surface. Furthermore, we observe a region of suppressed desorption close to the injection site, which behaves in a saw-tooth fashion as we increase the injection bias voltage. We show that each time the suppression region resets itself back to a minimum value, a new surface state appears as measured withscanning tunnelling spectroscopy (STS). We develop a theoretical model to link themagnitude of the supression region to the surface band structure, based on the coherent expansion of an electron wavepacket. By fitting the model to the experimental data we extract a coherent lifetime of 10 fs. We conclude that electron (and hole) transport is a two-step process: (1) ultrafast coherent inflation, followed by (2) incoherent diffusion.The second set of results presented here are aimed at understanding step (iii) ofthe nonlocal process, i.e. the molecular manipulation itself. To address this we perform extensive studies where we inject electrons (and holes) directly into a single target molecule on the surface and into the silicon adatoms themselves and look at the current and the voltage dependence of the manipulation rate. At low temperature, we compare directly the rate of desorption of toluene molecules to the rate of silicon adatom hopping. This, combined with a comparison of the voltage dependence of the manipulation rate with STS spectra acquired on top of a molecule and on top of a clean adatom, allows us to show that manipulation is in fact mediated by the underlying silicon surface. Injecting the tunnelling current into the molecule simply enhances this effect. Finally, the current dependence for hole injections into toluene molecules allows us to observe suppressed desorption rate at higher currents - the opposite effect of what is expected. By presenting the desorption rates as a function of tip height, we construct a simple model where we introduce a second decay channel for the excited state of the molecule: through the tip. This allows us to obtain an estimate of the tip-molecule separation during a `passive' scan of 0.4 A.In the context of all this, we discuss step (i), the injection process, and propose future experiments that will allow us to: gain better understanding of the injection process and measure the absolute reaction cross-section; extend the nonlocal manipulation to new surfaces, e.g. graphene, and employ this technique in order to observe relativistic quantum mechanics phenomena; and obtain quantitative information about some principal surface science properties on the nanoscale, like mobility.

Date of Award	23 Nov 2016
Original language	English
Awarding Institution	University of Bath
Supervisor	Peter Sloan (Supervisor)