Organically templated metal oxides have been the focus of intense interest for several decades owing to a host of technologically intriguing properties that they can exhibit.  Most of these are produced through exploratory reactions. Unfortunately, most of these reactions are deemed ‘unsuccessful’ in that they do not result in the formation of a new unique phase.  Such reactions are never reported in the literature, and there is no forum for collecting them, nor a means to fully extract chemical information from them.  Nevertheless, these "dark reactions" are valuable because they define the bounds on the reaction conditions needed to successfully synthesize a product. A searchable online repository for reaction data has been created that enables better management, sharing, and utilization of these dark reactions. Progress in using the "dark reaction" dataset to accelerate exploratory synthesis by using machine learning will also be discussed. Using this dataset, standard cheminformatics techniques are used to compute derived properties, and then trained a decision-tree algorithm to predict the success of new reactions. Preliminary results in experimentally validating these predictions will be presented.<br/>

Dark reactions project: a machine learning approach to materials discovery

249th National Meeting (2015)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.