Skip navigation
Please use this identifier to cite or link to this item:
Title: Part I. Engaging Iminium Ions and Radicals in Nickel Catalysis. Part II. Predictive Reaction Modeling Using Machine Learning.
Authors: Ahneman, Derek
Advisors: Doyle, Abigail G
Contributors: Chemistry Department
Keywords: cheminformatics
C-H functionalization
machine learning
Subjects: Organic chemistry
Issue Date: 2017
Publisher: Princeton, NJ : Princeton University
Abstract: Chapter 1 outlines the development of an enantioselective arylation of quinolinium ions mediated by nickel catalysis. This method yields 2-aryl and 2-heteroaryl-1,2-dihydroquinolines with moderate to high levels of enantioselectivity and demonstrates the utility of the Ni-iminium activation mode for asymmetric catalysis. Key to the development of this method was the identification of a Ni(II) precatalyst that can be activated under mild conditions. Chapter 2 chronicles the advent and development of Type II metallaphotoredox catalysis, defined herein as the coupling of an oxidatively generated radical with an oxidative addition substrate. Specifically, we developed a direct cross-coupling of amines or carboxylic acids with aryl halides using a nickel-photoredox dual catalyst system. These mild cross coupling protocols provide direct access to benzylic amines from inexpensive and readily available starting materials without the need for prefunctionalization. Finally, a second generation C–H, C–X coupling was developed using cyclic N-aryl amines. Chapter 3 describes the development of an automated procedure for the calculation and extraction of molecular, atomic, and vibrational descriptors. This process was used to explore the Buchwald–Hartwig reaction using heterocyclic fragment additives. Machine learning models were evaluated for their ability to predict reaction outcomes in high-dimensional reaction space. A random forest approach proved optimal, capable of predicting yield for an independent test set within a RMSE of 7.8% and with an R^2 value of 0.96. Finally, a random forest model predicted the performance of out-of-sample additives with an R^2 value of 0.91.
Alternate format: The Mudd Manuscript Library retains one bound copy of each dissertation. Search for these copies in the library's main catalog:
Type of Material: Academic dissertations (Ph.D.)
Language: en
Appears in Collections:Chemistry

Files in This Item:
This content is embargoed until 2018-06-09. For more information contact the Mudd Manuscript Library.

Items in Dataspace are protected by copyright, with all rights reserved, unless otherwise indicated.