(727a) Identification of Catalyst Descriptors for Oxidative Coupling of Methane

Oxidative coupling of methane (OCM) has attracted substantial interest due to the potential gains that direct conversion of methane into higher hydrocarbons, mainly ethane and ethylene, can provide. However, OCM is challenged by low C₂ yields due to the undesired CO_x formation. Given the large number of experimentally measured OCM performance data on various metal-oxide catalysts there have been multiple attempts to use data science tools to analyze past performance data and predict new catalyst compositions.^1-3 These past attempts, however, have been unsuccessful in identifying the key catalyst features that are responsible for good OCM performance; possibly, because these features are not readily accessible in available databases and were not included in the training data for model development.

To this end, we examine a smaller set of mixed metal-oxides and follow a hypothesis-driven research approach to identify relevant surface-adsorbate interactions that can correctly predict OCM performance trends. We focus our investigation initially on pure metal oxides and doped metal oxides (with alkali, alkaline earth metals). For selected oxide surfaces, we use DFT to construct a partial potential energy diagram for critical OCM reaction steps, including hydrogen abstraction from methane and oxygen vacancy formation. We then test for correlations between the energetics of these steps with simpler energy descriptors, for example, the hydrogen or methyl binding energy. Moreover, the Bader charge partitioning method is employed to estimate the oxidation state of participating oxygen species from which the relationship between oxygen oxidation state and hydrogen abstraction energy is assessed. By selecting a well-defined set of doped metal oxides and we can selectively tune the surface catalytic properties and link the experimentally observed performance trends to intrinsic properties of the catalyst material. Ultimately, these properties will be tested as OCM performance descriptors for a larger set of mixed metal oxides and included in a machine learning model to predict improved catalysts for OCM.

References

(1) Zavyalova, U.; Holena, M.; Schlögl, R.; Baerns, M.; ChemCatChem 3 (2011) 1935–1947.

(2) Kondratenko, E. V; Schlüter, M.; Baerns, M.; Linke, D.; Holena, M. Catal. Sci. Technol. 5 (2015) 1668–1677.

(3) Takahashi, K.; Miyazato, I.; Nishimura, S.; Ohyama, J. ChemCatChem 10 (2018) 3223–3228.