Understanding complex Pd-catalysed reaction systems through automated chemistry, data analysis and mechanistic studies

Automated chemical synthesis using robotics brings the benefits of reproducibility and
the potential to screen reactions in many conditions, increasing opportunities for
optimisation of reaction discoveries. Batch screening, using the Chemspeed robotic
platform, and flow screening methodologies were used to investigate a complex Pd-catalysed
reaction of 2-bromo-N-phenylbenzamide involving the formation of
phenanthridones. This provides an excellent example of a reaction system that can be
tested via automation as the substrate forms many products, allowing the study of
product distributions in different conditions. Increased understanding of side product
formation and catalytic cycles can help manage safety concerns in the synthesis of a
pharmaceutical drug molecule and the reaction system chosen here forms
pharmaceutically-validated privileged phenanthridinone structures. The LC-MS data obtained from batch screening were analysed using principal component analysis and regression decision trees, which showed that solvents have the largest effect on product distribution. Interpretation of the product clusters observed in
different solvents provided understanding of the mechanistic picture. DMF and propylene
carbonate (green alternative solvent) was found to give the highest conversion (≈60%)
to phenanthridinone product and n-Bu2O, the lowest conversion to phenanthridinone
product, whereas side products were seen in MeCN. After validation of the robotic batch screening method, comparison with an established flow approach showed the maximum conversion to the phenanthridinone
product was similar in DMF, but lower in batch than flow for MeCN.