We have developed a new method for modeling the electronic features of molecules that uses confined electrons on the surface of copper. Our approach can measure charge-density, bond order, bond length, and other molecular properties. Using a scanning tunneling microscope (STM) to perform atomic manipulation, we present techniques that confine electrons into quantum corrals and demonstrate real molecular properties.
Before constructing synthetic molecules, we refined the computer simulations used to optimize our designs by amending the dispersion relation to account for surface state electrons scattering to bulk states, and we improved the scattering phase by minimizing the error between simulations and experiment. Comparing our results with STM measurements showed an improvement.
Next, we demonstrate methods of constructing synthetic molecules and show they agree with real molecules and density functional theory (DFT). We start by building a hydrogen atom, which consists of carbon monoxide molecules dosed onto the surface of Cu(111). To establish that our synthetic hydrogen atom represents the properties of real hydrogen, we create a hydrogen molecule and show that the bonding and anti-bond states have the correct relative energies. Next, we show that we can also model planar molecules where the valence electrons are π-bonds. We construct benzene, anthracene, and phenanthrene and show that the charge densities are correct. To conclusively show that our molecules represent real ones, we extract the correct bond orders and bond lengths.
Finally, we propose and show initial results for modeling a single molecule transistor and a single molecule strain sensor. These techniques provide a powerful tool for exploring the properties of molecular devices that do not exist. We model the transistor using artificial benzene and additional rectangular quantum corrals on the outside, which simulate electrical leads. To simulate charge transfer, we measure coherence between the rectangle corrals. We use these same techniques to model a strain sensor.
This dissertation demonstrates that we have developed the techniques for a new method of modeling molecules. We have also demonstrated promising initial results that show the usefulness of these techniques by constructing models of transistors and sensors that are not yet possible with actual molecules.