Anti-sintering MgAl2O4 supported Pt-Ge nanoparticles for propane dehydrogenation: Catalytic insights and machine-learning aided performance analysis

In the present study, the catalytic performance of Pt nanoparticles in Propane Dehydrogenation (PDH) over two different supports (Commercial Al2O3 and MgAl2O4) was evaluated using experimental methods combined with Machine Learning. Different characterization techniques, including HAADF-STEM, H2-TPR, NH3-TPD, C3H6-TPD, XPS, BET, CO-chemisorption, CO-DRIFT, and TPO, were used to unravel the correlation between catalytic performance of the Pt nanoparticles with the promoter and the supports employed during PDH. Firstly, experimental analyses indicate that Ge modifies the geometric and electronic properties of Pt as the active metal in the reaction. Moreover, using MgAl2O4 instead of commercial Al2O3 support boosted the performance of Pt-Ge nanoparticles due to its anti-sintering nature and fewer acidic centers. Finally, the influence of operation conditions on the reaction rate of the best sample was investigated with machine learning. The gradient boosting tree model was employed to learn from relevant variables, including temperature, partial pressure of propane, and partial pressure of hydrogen and propylene in the feed. The results reveal the importance of the parameters on the reaction rate as Temperature > P(C3H8) > P(C3H6) > P(H2). Even though hydrogen has the least effect on the reaction rate, it is essential to co-feed hydrogen to suppress coke formation in PDH.