Date Thesis Awarded
5-2023
Access Type
Honors Thesis -- Access Restricted On-Campus Only
Degree Name
Bachelors of Science (BS)
Department
Chemistry
Advisor
Kristin Wustholz
Committee Members
Nathan Kidwell
Tyler Meldrum
Ron Smith
Abstract
Despite the large-scale adoption of data science in neighboring fields and plethora of opportunities available, chemistry has been relatively slow to incorporate it in research. To expand the intersection of chemistry and data science, computational algorithms are developed to improve photovoltaic design and designing a novel method of multicolor, nanoscale optical imaging.
Dye-sensitized photocatalysis is a promising solar technology to meet the ever-growing energy needs. Yet, its efficiency is hindered by a lack of understanding about the distribution of kinetics underlying electron transfer. By removing the effects of ensemble averaging and molecular aggregation, the first part of this thesis untangles the factors that govern the dispersive electron transfer kinetics of eosin-sensitized TiO2. Doing so required the development of a change point detection (CPD) algorithm and a high-throughput Monte Carlo simulation screening program. Combined, these elucidated the kinetic model and rates associated with electron transfer at the Eosin Y-TiO2 interface and laid the groundwork for analyzing other emitters in a statistically robust manner.
Taking advantage of CPD, the second part of this thesis showcases several proof-of-concept studies in developing Blinking-Based Multiplexing, a novel technique for differentiating between spectrally overlapped emitters. By implementing a machine learning algorithm to classify emitters by their emission patterns over time, this research created a rapid, accurate, and readily generalizable method for multicolor super-resolution imaging. After surveying a variety of machine learning algorithms, logistic regression appeared best for differentiating between two spectrally overlapped emitters, routinely achieving 95% accuracy across different experimental conditions.
Recommended Citation
Hoy, Grayson, "Development of Data Science Tools for Photovoltaic Design and Super-Resolution Imaging" (2023). Undergraduate Honors Theses. William & Mary. Paper 2023.
https://scholarworks.wm.edu/honorstheses/2023
