Computational history of chemical reactions

The chemical community devotes most of its efforts to synthetic chemistry, therefore knowledge about reactants, catalysts, solvents and several other related aspects of chemical reactions is of important relevance. Part of this knowledge is its history that involves determining the aspects that have shaped chemical reactions to their current state; which are tasks for the history of chemistry. However, analysing the chemical reactions that have been reported in the scientific literature is not any more a subject of the conventional history of chemistry, for the number of substances and reactions grows exponentially. Here we show that a computational approach to the history of chemical reactions sheds light on the patterns behind the development and use of substances and reaction conditions. We explored the more than 45 million reactions gathered in Reaxys database by modelling them as a network through a bipartite hypergraph model of a chemical reaction. We came across with historical patterns for substances, types of substances, catalysts, solvents, temperatures and pressures of those reactions. It is found that chemists have traditionally used few reactants to produce many different substances. In such synthesis more combinations of about four chemical elements are explored. Despite the exponential growth of substances and reactions, little variation of catalysts, solvents, and reactants is observed throughout time. Regarding reaction conditions, the vast majority of reactions fall into a narrow domain of temperature and pressure, namely normal conditions. We also found and quantified the effect of world wars (WWs) upon chemical novelty during war periods. WW1 took production of new substances and reactions back around 30 years and WW2 around 15. We anticipate this study and especially its methodological approach to be the starting point for the computational history of chemical reactivity, where social and economical contexts are integrated.

Joint work: Eugenio J. Llanos, Wilmer Leal, Guillermo Restrepo, Peter Stadler