A closed-loop framework for descriptor-empowered exploration of high-activity Co-based quaternary molten alloy...

Methane (CH₄) pyrolysis via molten media offers an attractive route to CO₂-free hydrogen production with high-value carbon as a co-product. However, the rational design of multicomponent molten catalysts remains challenging due to the dynamic and disordered nature of active sites in the molten state. Here, we present a closed-loop framework that integrates potential catalyst exploration by descriptor data mining, CH₄ pyrolysis testing, and activity mechanism analysis to identify high-activity catalysts for CH₄ pyrolysis. Using this approach, Bi₇₀Ni₁₈Cu₆Co₆ is identified as a promising catalyst, with experimental validation demonstrating that it can achieve 40.3% CH₄ conversion at a moderate 1000 °C with a relatively low activation energy (197.3 kJ mol^-1), outperforming Bi₇₃Ni₂₇ (33.3%) and Bi₇₆Ni₁₈Cu₆ (37.4%) molten media under the same experimental conditions. The enhancement stems from a synergistic multi-solute effect, whereby the co-dissolution of Ni, Cu, and Co in the low-surface-tension Bi solvent increases the density of exposed active sites and promotes direct interaction with CH₄ molecules. In particular, introducing Co downshifts the LUMO energy level of the CH₄ molecular orbital (∼1.3 eV), facilitating charge antibonding occupation, weakening C-H bonds, and thereby accelerating CH₄ dissociation. These findings provide deep insights into Co-based multi-elementary molten catalysts and chart a clear path toward the rational design of high-performance multicomponent molten catalysts for CH₄ pyrolysis.