Predictive modeling in homogeneous catalysis: a tutorial

Abstract

Predictive modeling has become a practical research tool in homogeneous catalysis. It can help to pinpoint ‘good regions’ in the catalyst space, narrowing the search for the optimal catalyst for a given reaction. Just like any other new idea, in silico catalyst optimization is accepted by some researchers and met with skepticism by others. The basic requirements for good predictive models are a reliable set of initial experimental data, a method for generating and testing virtual catalyst libraries, and robust validation protocols. Once you have these, the key task is translating the catalysis problems into something that a computer can understand. In this tutorial review we explain in simple terms what predictive modeling actually is, why and when should one use it, and how it can be implemented.