Computational Organic Chemistry Through Data Driven and Quantum Chemical Methods

Computer-aided calculations are frequently used for modeling chemical reactions. Both predictive and quantum mechanical models give insight into reaction mechanisms and stereochemical outcomes. C–H borylation reactions offer a synthetic route toward late-stage functionalization (LSF) of advanced natural product intermediates and bioactive medicinal compounds. Borylation reactions allow for the direct functionalization of C–H bonds into C-B bonds via boron functional handles, such as a boronic ester. A commonly used reaction with boronic esters is the Suzuki-Miyaura cross-coupling, which readily forms new carbon-carbon bonds. Despite its utility, there have been obstacles in correctly predicting the site of borylation prior to performing these reactions. Site-selectivity prediction models are increasingly gaining popularity in the organic chemistry sphere and have been used to address this hurdle. This work will cover our random forest model developed to predict site-selective borylations as well as provide insights gathered from SHAP (SHapley Additive exPlanations) analysis and t-SNE modeling of our dataset. This dissertation will include reaction mechanism studies using quantum mechanical methods for reactions involving borylated olefin systems. The first study focuses upon reactivity and stability trends when synthesizing boron-containing bicyclo[2.2.2]octenes via [4+2] cycloadditions. A second study involves the formation of δ-silyl-homoallylic alcohols via allylboration with α-silyl-substituted boron reagents and aldehydes. Computational investigations illuminate stereochemical outcomes based upon steric hindrance and unique hydrogen bonding amongst the reactants within the transition states. Further computational studies were conducted to probe mechanisms for oxidations that use Bobbitt’s salt as a chemical oxidant. The results were used to formulate probable reaction pathways and mechanisms for the experimental formation of disulfides from thiols as well as diazines from hydrazides. These studies provided unique insights into chemical reactivity of Bobbitt’s salt with sulfur and nitrogen-containing systems. Finally, simulated sunlight exposure experiments involving terrestrial dissolved organic matter (tDOM) are analyzed using the Neo4j graph database via a temporal graph network. This study helped form a better understanding of the organic reactions that fluvial tDOM undergoes, thus improving our understanding of tDOM’s life cycle as it travels into contributing waterways, and eventually into the ocean, where it makes the bulk of oceanic carbon stores.