A chemistry‐inspired neural network kinetic model for oxidative coupling of methane from high‐throughput data. Issue 6 (22nd January 2022)

• Record Type: Journal Article
• Authors: Chen, Kexin; Tian, Huijie; Li, Bowen; Rangarajan, Srinivas
• Is Part Of: AIChE journal. Volume 68:Issue 6(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-01-22
• Subjects: data science -- high‐throughput experiments -- kinetic modeling -- machine learning -- oxidative coupling of methane; Chemical engineering -- Periodicals; Génie chimique -- Périodiques; 660.28
• DOI: 10.1002/aic.17584
• Languages: English
• ISSNs: 0001-1541

Abstract

Abstract: A neural network kinetic model is developed for oxidative coupling of methane (OCM). The model is designed in cognizance of the underlying chemistry and associated reactor balance equations and trained on publicly available high throughput experimental data spanning a large material space of supported mixed metal oxide catalysts. The resultant model is then used to evaluate one of the most popular catalysts for OCM, viz. MnNa2 WO4 /SiO2, to understand the reaction kinetics and sensitivity of the catalyst to changing different components of the catalyst. The predicted activation barrier for methane conversion is 251 kJ mol −1, and the rate r ∼ CH 4 0.7 O 2 0.6 . Furthermore, the reference catalyst is local optimal as small changes to its composition, for example, by changing the individual metals or the support, did not improve (or often substantially reduced) methane consumption or the C2 formation rate.