Bayesian Optimization of Solvent-Free Thermal Amidation via Reactive Extrusion

A solvent-free thermo-mechanochemical synthesis of amides from carboxylic acids and amines, performed in an extruder, in the absence of coupling agents, additives, or catalyst is reported. A Bayesian optimization (BO) approach, incorporating an active and soft constraint, was developed, able to handle holistically various parameters (temperature, reaction time, excess of starting materials) while targeting a high conversion and sustainability. The methodology proved to be straightforward, high-yielding, sustainable, and easily adaptable to various substrates with a reduced number of experiments. It was applied to a variety of substrates, providing an easy and efficient preparation method of amides, including bioactive macamides and the active pharmaceutical ingredient (API) moclobemide. Optimized conditions facilitated the product isolation, some of them were obtained essentially pure from the extruder, others purified by extraction or crystallization. Notably, moclobemide could be synthesized at larger scale and isolated in the complete absence of organic solvent.