Describing the electronic structure of molecules on metal surfaces

Accurately describing the electronic structure of molecules on metal surfaces is key to correctly modeling their surface-enhanced properties. These properties are the basis for a variety of topics in chemistry, such as single molecule spectroscopy and organic photovoltaic systems. In fact, the most recent Nobel Prize in Chemistry was awarded for work in the field of single molecule fluorescence. While single-molecule uorescence is now widely used within both the chemical and biochemical communities, its spectroscopic signal gives very little information about the structure and identity of the fluorophore. Surface enhanced Raman spectroscopy (SERS), on the other hand, can be used to uniquely identify a molecule as well detect the presence of a known scatterer. Raman differs from uorescence, as its the result of the inelastic scattering of photons by a molecule rather than an absorption process. These scattered photons contain information about the vibrational and rotational states within the molecule, similar to IR spectroscopic techniques. However, the Raman signal from a single molecule is very weak. The mechanisms behind SERS provide sufficient enhancement to enable single molecule detection and identification. Modeling SERS and other surface-enhanced properties is difficult due to the complex interactions between the molecule and the metal surface. In order to accurately describe how these interactions impact the electronic structure, we require first-principles based methods. Density functional theory (DFT) remains the go-to method for simulations of large systems thanks to its balance between accuracy and computational complexity. However, one encounters certain failures within DFT that limit its application to accurately describing the interactions between molecules and metal surfaces. In principle, DFT is an exact method if one knows the correct exchange-correlation (XC) potential. In practice, this potential is only an approximation determined by an XC functional. Many XC functionals exist and the accuracy of a DFT calculation is highly dependent on the choice of XC. Recently, a new class of XC functionals called long-range corrected (LC) functionals have been developed that show significant improvement to the traditional failures of DFT. Of particular interest is their ability to be 'tuned' in order to enforce properties that the exact XC functional would have. In this dissertation, we present the importance of using LC functionals when describing the electronic structure of molecules on metal surfaces using DFT. We first demonstrate how LC functionals improve the description of the energy gap between the frontier orbitals for a set of substituted pyridines on a small silver cluster. This allows for a better prediction to the magnitude of the SERS enhancement. While DFT is capable of describing 'large' systems on the order of hundreds of atoms, realistically sized nanoparticles with dimensions on the order of 1 to 100 nm can contain between 300 and 10,000,000 atoms, making them computationally intractable even for DFT. In order to go beyond small metal clusters, we have developed a hybrid model that combines a quantum mechanical description of a molecules using density functional theory (DFT) with a classical atomistic electrodynamics model of the metal system. We present here a new implementation of the discrete interaction model/quantum mechanical (DIM/QM) method within the NWChem computational package. We demonstrate that by combining DIM/QM with the tuning of LC functionals, we can accurately describe the changes in electronic structure seen when molecules approach a metal surface at a significantly reduced computational cost compared to other methods. These changes are important to capture in a metal-molecule system, as they significantly alter the molecule's optical properties. In addition, we have made several improvements to the underlying DIM/QM algorithm which decrease the computational cost of DIM/QM by ≈ 30%. Furthermore, we extend DIM to account for experimentally observed changes in the optical properties of metal nanoparticles with dimensions less than < 10 nm. Finally, we introduce a novel stand-alone program used to rapidly generate roughened sphere and spheroidal geometries. In collaboration with an experimental group, we demonstrate that these roughened nanoparticles provide a more accurate description to the optical properties of irregularly shaped aerosol particles. The tools and methods presented here will further our understanding of metal-molecule interactions of large systems, such as those in the fields of single molecule electronics, plexitonics, and surface-enhanced Raman scattering.