The direct conversion of methane to value-added chemicals has not yet been fully realized as of today, but is considered a game-changer, if a suitable catalyst can be discovered. Computational catalyst screening based on energy descriptors and linear scaling relationships to estimate kinetic input parameters to microkinetic models for the rapid prediction of activity and selectivity trends among heterogeneous transition metal catalysts has gained much popularity in recent years. However, in its standard implementation, which makes assumptions based on known catalysts, this strategy is not well-suited to identify truly novel catalysts for processes that operate under nontraditional reaction conditions and are affected by other unidentified factors. The need for such assumptions and simplifications arose from the limited availability of computing resources as well as limited human imagination in higher dimensions. Thus, a minimal set of descriptors, usually no more than two, is used to ease visualization and data analysis by humans.<br/>Here, we will present a comprehensive catalyst design effort for active and selective methane upgrade processes by using density functional theory (DFT), kinetic Monte Carlo simulations (kMC), microkinetic modeling (MKM), and ,achine learning techniques (ML), as well as experimental verification. We elaborate on the predictive powers of each technique for the multiple time and length scales and aspects of a reaction, and argue that successful catalyst design must rely on the simultaneous analysis of available information from sources (theory, experiment, literature). Adopting techniques from the increasingly popular field of machine learning is a potential path forward to realize such comprehensive data analysis.<br/>

Computational design of active and selective catalyst for methane conversion to value-added products

255th National Meeting (2018)

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