Oxidative coupling methane (OCM) is challenging reaction as it operates at high temperature (&gt;700 ○C) resulting the production of COx becomes crucial byproduct. Furthermore, its reaction route consists of gas phase and surface reactions with complicated-stepwise procedure. Thus, controlling reaction has been attempted via catalyst in order to obtain desirable C2 production. However, catalyst design for OCM reaction has been stagnated for decades because of its complexity of reaction. On the other hand, data science approach appears to be a powerful approach to unveil the undiscovered properties of OCM reaction.Here, catalysts informatics is proposed to unveil the details mechanism of OCM reaction via experiment.In particular, multi-output machine learning successfully predicts the selectivity of CO, CO2, C2H4 , C2H6, C2, and CH4 conversion instantaneously within the X-Na2WO4/SiO2(X:(none), Mn, Cu, and Ti) catalysts where the change of each selectivity against experimental conditions are fully unveiled by interpolation filling. The result reveals the not only the effect of X in X-Na2WO4/SiO2 catalysts, but also the trade-off relation between C2H6 and CO2 selectivity, providing the guidance of both catalyst design and optimized experimental conditions. Additionally, it is discovered that the reaction network can be represented via machine learning based on the trend in experimental data. Therefore, data science technique advances the study of OCM reaction, indicating to be a promising approach for catalyst research.

Drawing reaction network and details mechanism of methane oxidation via machine learning

ACS Spring Meeting 2021

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