General Reaction

Machines are really good at doing the same thing over and over again. In contrast, chemists like to modify the recipes they use every time they make a different molecule. This tradition of customizing recipes is especially prevalent when chemists make the difficult bonds between carbon atoms, and because such carbon-carbon bonds are the main types of bonds in small molecules, it has been very challenging to automate this traditionally highly customized process of small molecule synthesis. We thus launched a mission to find more general conditions for a certain type of carbon-carbon bond formation. To achieve this, we created a frontier “closed-loop” discovery engine leveraging MML’s automated block chemistry in concert with artificial intelligence and automated reaction characterization. In a creative gambit, the AI was empowered to make lots of bad suggestions so that ultimately it could make some really good ones. And this gambit paid off. This mission ultimately yielded a set of highly general conditions for forming an important class of carbon-carbon bonds. This breakthrough also enabled a major step forward for automated iterative carbon-carbon bond formation.

Publication:

Science – Closed-loop optimization of general reaction conditions for heteroaryl Suzuki-Miyaura coupling

Highlighted:

Chemical & Engineering News – AI-based robot uncovers general reaction conditions for Suzuki-Miyaura coupling

U of I News – Artificial intelligence and molecule machine join forces to generalize automated chemistry