Work

Integrating Theory–Experiment Active Learning and Mechanistic Insights for High-Entropy Alloy Design Toward Efficient Alkaline Hydrogen Evolution

Year: 2026

Type: preprint

Abstract: High-entropy alloys (HEAs) offer diverse local coordination environments for modulating hydrogen adsorption and water activation in alkaline hydrogen evolution, but their highdimensional compositional space remains difficult to explore efficiently. Theory-experiment closed loops can accelerate this search, yet theoretical prior and candidate sampling are commonly coupled, obscuring their individual contributions. Here, we developed a theory-experiment activelearning (TEA) framework that independently varies theoretical priors and candidate sampling strategies for AuCuPdPtIr HEAs. Under identical experimental throughput, the theoretical prior provided transferable performance trends across composition space, whereas the proposed adaptive sampling strategy broadened effective compositional coverage. After three iterations involving 120 compositions, the framework identified Au16Cu5Pd35Pt35Ir9, which required only 16 mV overpotential to reach 10 mA cm-2 in 1 M KOH. An anion-exchange membrane water electrolysis (AEMWE) cell using this catalyst as the cathode reached 1 A cm-2 at 1.647 V and 60 °C and operated stably for 1000 h. Model interpretation, operando infrared spectroscopy, machine-learning-potential molecular dynamics (MLP-MD) simulations, and charge-density analysis further associated Ir/Cu-related local environments with a weaker interfacial hydrogenbond network and more Volmer-like water configurations. This work connects controlled module comparison, data-efficient compositional optimization, and interfacial-water mechanisms, providing a general strategy for complex multicomponent electrocatalysts.

Source: ChemRxiv

DOI: 10.26434/chemrxiv.15008410/v1