Enantioselective Synthesis of Sterically Hindered α-Aryl Stereocentres in N-Heterocycles

Pd-mediated decarboxylative catalysis has become a mild, but powerful tool for the asymmetric synthesis of stereocentres adjacent to a carbonyl group. In 2004, Both Tunge and Stoltz reported the first decarboxylative asymmetric allylic alkylation (DAAA), generating quaternary stereocentres in high enantioselectivities. Generally, the identity of the α-substituent in the formed stereocenter remained a sterically small group, such as a methyl or a benzyl substituent. Our group has focused on generating quaternary stereocentres bearing highly sterically hindered α-aryl substituents, where, in previous reports, we have been successful in obtaining exceptionally high enantioselectivities of up to > 99 % ee for a range of heterocyclic and non-heterocyclic substrates. Another form of decarboxylative catalysis that has been of interest to our group is the Pd-catalysed decarboxylative asymmetric protonation (DAP), which has been employed to generate tertiary stereocentres in high enantioselectivities. First reported by Muzart in 1992 via a chiral alcohol-mediated variant, it was further developed by Stoltz in 2006 as an extension to their previous DAAA work. Similar to our own DAAA work, we have been successful in extending DAP methodology to substrates containing bulky α-aryl substituents. Our group has reported both chiral amino alcohol-mediated DAP and chiral P,N ligand DAP methodology for the synthesis of α-aryl tertiary stereocentres from a range of heterocyclic and non-heterocyclic substrates, obtaining up to 95 % ee. The aim of this PhD project has been to further extend the scope of both DAAA and DAP methodology to two more N-heterocyclic substrates: α-aryl lactams and α-aryl 4-piperidones. The initial work in this project involved the synthesis a range of α-aryl substrates, the key step being a Pb-mediated α-arylation. We successfully synthesised 14 novel α-aryl lactam substrates, 11 α-aryl variation and 3 N-protecting group variations. We also successfully synthesised 13 novel α-aryl 4-piperidone substrates, all which are α-aryl variations. We next applied these novel substrates in DAAA and DAP catalysis. A substrate scope of 11 examples of DAAA with 6-membered lactam substrates demonstrated enantioselectivities of up to 82 % ee. Optimisation of DAAA with a 5-membered lactam substrate resulted in lowered enantioselectivity, with ees up to only 20 %. With the α-aryl 4-piperidone substrates, a substrate scope of 13 examples obtained excellent enantioselectivities of up to 99.9 % ee, with products bearing di- and mono-ortho substitution patterns showing the highest levels of enantioselectivity. We next applied these novel substrates to chiral amino alcohol-mediated DAP methodology, where it was found that (1R, 2S)-(-)-ephedrine, in the absence of a chiral ligand, afforded high enantioselectivities of up to 95 % ee with lactam derived substrates and up to 90 % ee with 4-piperidone derived substrates. A substrate scope of 20 examples showed that products containing di-ortho-methoxy-substituted phenyls and naphthyl groups showed the highest ees, whereas products not bearing this substitution pattern gave lower levels of enantioselectivity. Finally, Chapter 6 describes the work undertaken in a 2 month placement in the University of Nottingham in collaboration with Prof. Simon Woodward, where we applied Machine Learning (ML) to the DAAA data that our group has collected throughout the years. We successfully expanded upon the system the Woodward group previously reported for Rh-catalysed 1,4 additions, generating a simple quantitative structure-property relationship between the enantioselectivity of the Pd-catalysed DAAA and Trost-type ligands/substrate structure. Using easily interpretable chemical features, ML models were developed with RMSEs as low as 8 %, through previously optimised algorithms and a small number of DAAA examples.