Oxidative Cracking\nof <i>n</i>‑Hexane\nto Light Olefins over VO<sub><i>x</i></sub>/MgO-γAl<sub>2</sub>O<sub>3</sub> Using Lattice Oxygen: Performance Evaluation\nand Machine Learning Modeli

This study investigates VO<sub><i>x</i></sub>/MgO-γAl<sub>2</sub>O<sub>3</sub> in the oxidative cracking\nof<i>n</i>-hexane to produce light olefins in the absence\nof gas-phase oxygen.\nThe catalysts were prepared with varying mass ratios of MgO/γAl<sub>2</sub>O<sub>3</sub>(1:2, 1:1, and 2:1), while the VO<sub><i>x</i></sub> loading was maintained at 10 wt %. Among the synthesized\ncatalysts, VO<sub><i>x</i></sub>/MgO-γAl<sub>2</sub>O<sub>3</sub> 1:1 showed superior catalytic activity, with 89.1% <i>n</i>-hexane conversion and 92.6% light olefin selectivity.\nIntroducing an appropriate amount of MgO enhanced the dispersion of\nVO<sub><i>x</i></sub> active species, balanced the acidity,\nand suppressed the oxidation of hydrocarbons. Additionally, a machine\nlearning model was developed to predict oxidative cracking products’\nyields. The model, based on 44 data points from this study and literature,\npredicted <i>n</i>-hexane conversion, olefin yield, carbon\noxide yield, methane yield, and paraffin yield using catalyst formulations,\ntemperature, and time as inputs. The model showed a high correlation\n(<i>R</i><sup>2</sup>) of 0.99 and RMSE values between 1.6\nand 8.5, highlighting its strong predictive capability.