Non-empirical tuning in DFT: improvements for modeling charge transport parameters in organic semiconductors

This dissertation is focused on modeling charge-transport in π-conjugated organic materials, which serve as the active materials in light-weight, flexible, organic photovoltaic cells, offering the potential for cheap, ubiquitous renewable energy. In particular, we used computational chemistry to gain insight into the fundamental processes of charge transport within organic semiconductors to derive an understanding of chemical and physical phenomena that can not be explained through experiment alone in order to further the performance of organic-based electronic devices. In order to accurately model the organic materials, a combined quantum-mechanical and classical approach is needed, with the ground and excited state electronic properties of isolated organic materials determined using DFT/TD-DFT as a first step and coupled with molecular dynamics and mechanics. This allows for an understanding of the molecular order and packing within nanoscale structures as well as the impact of the intermolecular interactions. However, standard DFT methods suffer from intrinsic errors resulting from approximations to the exchange-correlation potential that can be corrected using a simple non-empirically tuning procedure. We briefly review the electronic structure methods that we use and the non-empirically procedure for DFT that allows for a substantial improvement over standard DFT methods. We then discuss the main results of this research. In Chapter 3, we detail the understanding of the limitations in DFT (currently one of our main tools) and improvements that can be achieved through non-empirically tuning a specific DFT method for the system of study. We detail the dependence of the range-separation parameter used in long-range corrected hybrid functionals on both the size and degree of conjugation for a given system. We also demonstrate the effect that self-interaction corrections employed through range-separated hybrid functionals can have in describing thermodynamic and electronic properties for large, organic π-conjugated systems. In this study, we chose a property that critically depends on the degree of delocalization (i.e., torsion potentials) to correlate the degree of delocalization with the choice of a given method in order to understand how the self-interaction errors affects this property. These results are published in C Sutton et al. “Accurate Description of Torsion Potentials in Conjugated Polymers using Density Functionals with Reduced Self-interaction Error” Journal of Chemical Physics, 140, 054310, 2014. In Chapter 4, we discuss how non-empirically tuning DFT can be used to rigorously model electron transfer in single-molecule systems (i.e., organic mixed-valence systems), where we modeled the symmetry breaking and charge (de)localization in charge-transfer complexes compared with high-level methods. The results presented in Chapter 4 are published in C Sutton et al. “Towards a Robust Quantum-Chemical Description of Organic Mixed-Valence Systems” Journal of Physical Chemistry C, 118, 3925, 2014. In Chapter 5, we applied this method to interpret photoelectron spectroscopy spectra in order to elucidate the localized nature of a charge carrier in prototypical organic semiconductors; this understanding was then extended to quantify the relaxation energy in finite molecular clusters in the presence of an excess charge from a combined multi-layer quantum-mechanical/molecular-mechanical method. In Chapter 6, we determined the effect of choosing various DFT methods on the intermolecular electronic couplings and band structure calculations in organic molecules, which are published by C Sutton et al. in "Understanding the Density Functional Dependence of DFT-Calculated Electronic Couplings in Organic Semiconductors” Journal of Physical Chemistry Letters, 4, 919, 2013. Finally, conclusions and further considerations are discussed in Chapter 7.Ph.D