Chemists still analyse catalytic reactions using methods developed nearly a century ago, relying on complex equations and mathematical approximations that miss most of the information hidden in their experimental data. This matters because understanding exactly how a catalyst drives a reaction is the key to designing better ones—faster, cheaper, and greener. Current kinetic analysis techniques focus on crude "kinetic orders" and require researchers to derive rate laws by hand, discarding proposals that don't fit simplified models. The result is a slow, error-prone process that leaves rich data unexploited. The researchers plan to train machine learning models on experimental kinetic data, teaching them to extract all available information and automatically propose reaction mechanisms consistent with the data. The models will handle the most common types of catalytic reactions used in academic and industrial labs—those involving one or two substrates. If successful, the tools will streamline the development of synthetic methods in pharmaceuticals, agrochemicals, and other sectors that depend on catalytic synthesis. For a senior programme manager, the payoff is clear: faster, more rational catalyst design, less trial-and-error in the lab, and a direct route to greener chemical processes.
View original technical description
Understanding the mechanisms of catalytic organic reactions is essential for advancing the design of new catalysts, exploring novel modes of reactivity, and developing greener, more sustainable chemical processes. Mechanistic understanding provides invaluable insights, empowering chemists to enhance chemical systems and discover new reactions in a rational and informed manner. Kinetic analysis lies at the heart of mechanistic investigations, enabling researchers to test their hypotheses directly using experimental data, allowing them to discard inconsistent proposals. However, with a few recent exceptions, current kinetic analysis pipelines rely mostly on techniques developed nearly a century ago and require the derivation of complex rate law equations involving multiple mathematical approximations that limit their applicability. Furthermore, by focusing on "kinetic orders" of reagents and catalysts, these techniques often miss out on much of the rich information present in reaction kinetic profiles. In this work, we aim to harness the power of artificial intelligence to create a more robust and comprehensive approach to kinetic analysis. Specifically, we aim to apply machine learning to the challenge of creating a model capable of processing experimental kinetic data, extracting all kinetic information, and subsequently use this information to automatically propose one or more mechanisms that are compatible with the data. We aim to develop models able to tackle various types of catalytic reactions involving one and two substrates, covering the vast majority of reactions used in academic and industrial organic chemistry laboratories. This work will lead to the development of tools that will benefit scientists in both academia and industry streamlining the development of efficient synthetic methodologies in sectors such as pharmaceuticals, agrochemicals and others that rely on the synthesis of compounds through catalytic processes.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know