Post-ageing guided closed-loop discovery of multi-element alloy catalysts for automotive exhaust purification

Multi-element alloy catalysts exhibit tunable electronic structures and remarkable thermal stability, making them promising materials for automotive exhaust purification. However, most data-driven explorations have emphasised fresh activity, overlooking the post-ageing durability that governs real-world performance. Here, we have developed a closed-loop high-throughput discovery framework that employs post-ageing activity as the principal design index and integrates inverse analytical prediction to accelerate the development of durable high-entropy alloy catalysts. A total of 1493 catalysts were synthesised and automatically evaluated, yielding over one hundred compositions surpassing Pd benchmarks in low-temperature activity, total conversion, and durability. Mechanistic analyses revealed that the enhanced performance originates from cooperative sites formed between different elements-indicating synergistic adsorption behaviour beyond that of individual metals-and from synthesis conditions involving low temperatures and high alkalinity, which suppress the formation of mixed oxides with alumina and thereby optimise metal-support interactions. Furthermore, multi-component evaluations including low-reactivity hydrocarbons (i-C₅H₁₂) clarified the coupled redox behaviour between NO reduction and hydrocarbon oxidation, realistically reproducing actual TWC operation. Statistical validation demonstrated over twentyfold higher discovery efficiency than random exploration (p < 0.001). This study establishes a durability-aware, data-driven paradigm linking alloy design, process informatics, and machine learning toward practical, platinum-group metal-efficient automotive catalysts.