Skip to main content
eScholarship
Open Access Publications from the University of California

Development of C-C and C-N Bond Forming Reactions Utilizing Palladium Carbene Intermediates and Reaction Predictor: Using Machine Learning to Help Identify Unknown Side-Products in Organic Reactions

  • Author(s): Gutman, Eugene
  • Advisor(s): Van Vranken, David L
  • et al.
Abstract

The research described herein consists of two disparate areas of study. The first and largest portion describes the development of novel palladium-catalyzed C-C and C-N bond forming reactions. The second portion describes the development of the Reaction Predictor system and its application towards identification of reaction products and pathways.

Palladium-carbenes are important intermediates in many modern C-C and C-heteroatom bond forming reactions. Palladium-catalyzed carbenylative coupling reactions are analogous to carbonylative processes with carbon monoxide. Insertion of a cis X type ligand into the palladium-carbene can potentially generate a new stereogenic center, making these reactions worthy of study. Carbenylative insertions have been used to generate electrophilic η3- allylpalladium species which were trapped with nitrogen and carbon nucleophiles. This work describes the cyclization of η3-benzylpalladium species derived from palladium-catalyzed carbenylative insertion. This optimization and broad substrate scope of this reaction led to the synthesis of 1-arylindanes and 1-aryltetralins in high yields. Additionally, this reaction was used to prepare tetralone 2.30bl, a synthetic intermediate in the Curran synthesis of (±)-podophyllotoxin. The carbenylative cyclization also led us to pursue utilizing aliphatic N-tosylhydrazones as palladium-carbene precursors in other coupling reactions. It was realized that aliphatic N-tosylhydrazones with adjacent hydrogens can effectively participate in three-component palladium-catalyzed carbenylative cross-coupling reactions of vinyl iodides while avoiding β-hydride elimination.

Development of a palladium-catalyzed enantioselective carbene insertion into the N-H bond of aromatic heterocycles to obtain α-(N-indolyl)-α-arylesters and α-(N-carbazolyl)-α-arylesters, using α-diazo-α-arylacetates as palladium carbene precursors is also described. Aliphatic amines were also competent coupling partners in the reaction, affording biologically active piperdine derivatives in moderate yields. The reaction was applied towards the synthesis of a bioactive carbazole derivative in a concise manner.

In a separate project an inductive machine learning reaction prediction program called Reaction Predicator has been trained and applied towards identification of plausible reaction products in ESI spectra. The reaction predictor training set has been expanded by the addition of new reactions written in our lab. Over 800 transition metal based training reactions have been written. In addition, over 10,000 new complex training reactions have been written and added to the training set. The Reaction Predictor pathway search feature has been customized to match products to unknown m/z peaks in ESI spectra. Pathway search was applied towards unknown identification in palladium-catalyzed N-H insertion reactions.

Main Content
Current View