relation: http://d-scholarship-dev.library.pitt.edu/31533/
title: Rapid Computational Discovery of Pi-Conjugated Materials
creator: Kanal, Ilana Yocheved
description: The focus of this thesis is conjugated polymer properties for improved computational discovery of $\pi$-conjugated materials. Combination of these materials in differing orders alter the electronic structure and in tetramers, on average, an energy effect is seen. When expanded to hexamers, it became apparent that a more complicated effect exists that depends on block length and placement of that block or sequence within a hexamer. Sequence effects were applied both computationally and experimentally for combinations of benzothiadiazole and phenylene vinylene monomers to confirm importance of sequence in both solar cell performance as sequence can affect intrinsic and bulk properties orthogonally, such as HOMO-LUMO gap.     In addition to sequence study, inverse design of conjugated polymers from computed electronic structure properties demonstrate that while it is unreliable to predict polymer properties from the monomer properties alone, it is very reliable to make predictions from simple models. These models allow for better polymer property predictions without costly polymer calculations.    A large scale computational investigation assessing the utility of common classical force fields for computational screening of low energy conformers provided us with insight for the most reliable methods to use when screening molecules. Using statistical analyses on the energies of up to 250 diverse conformers of 700 different molecular structures, we find that energies and geometries from widely-used classical force fields show poor energy correlation with semiempirical and DFT energies calculated at PM7 geometries. In contrast, semiempirical (PM7) energies show better correlation with DFT calculations. With these results, we make recommendations for more reliably carrying out conformer screening.     Sequence effect, models for polymer predictions and assessment of classical force field methods for low energy conformer predictions are combined to produce our genetic algorithm to rapidly, computationally select materials. Optimization of our genetic algorithm shows that with relatively few calculations, millions of molecules can be screened with a significant speedup compared with brute force calculation of those same molecules.
date: 2017-09-24
type: University of Pittsburgh ETD
type: PeerReviewed
format: application/pdf
language: en
identifier: http://d-scholarship-dev.library.pitt.edu/31533/1/Kanal_dissertation.pdf
identifier:   Kanal, Ilana Yocheved  (2017) Rapid Computational Discovery of Pi-Conjugated Materials.  Doctoral Dissertation, University of Pittsburgh.    (Unpublished)