Machine learning insights into the production and characteristics of carbon nanotubes from methane catalytic decomposition

The sustainability of methane catalytic decomposition is significantly enhanced by the production of high-quality value-added carbon products such as carbon nanotubes（CNTs）. Understanding the production yields and properties of CNTs is crucial for improving process feasibility and sustainability. This study employs machine learning technique to develop and analyze predictive models for the carbon yield and mean diameter of CNTs produced through methane catalytic decomposition. Utilizing comprehensive datasets from various experimental studies, the models incorporate variables related to catalyst composition, catalyst preparation, and operational parameters. Both models achieved high predictive accuracy, with R2 values exceeding 0.90. Notably, the reduction time during catalyst preparation was found to critically influence carbon yield, evidenced by a permutation importance value of 39.62%.Additionally, the use of Mo as a catalytic metal was observed to significantly reduce the diameter of produced CNTs. These findings highlight the need for future machine learning and simulation studies to include catalyst reduction parameters, thereby enhancing predictive accuracy and deepening process insights. This research provides strategic guidance for optimizing methane catalytic decomposition to produce enhanced CNTs, aligning with sustainability goals.