Automatic Discovery and Optimal Generation of Amorphous High-Entropy Electrocatalysts.

Amorphous materials are ubiquitous in nature and are widely used for many industrial applications, including catalysis, energy storage, and environmental science. However, significant challenges remain in designing and optimizing amorphous high-entropy materials because of the lack of well-defined structure-activity relationships. Here, we use synthesis systems to discover and optimize amorphous high-entropy oxyhydroxide electrocatalysts within the entire design space for the alkaline oxygen evolution reaction. Amorphous high-entropy electrocatalysts are derived from ultrathin 2D coordination polymers composed of six nonprecious metal elements that were selected from top 16 candidate metal elements involved in oxygen evolution reaction (OER)-related literature searching, which can then be transformed into amorphous oxyhydroxides. Leveraging machine learning (ML) techniques, we establish a composition-activity relationship and thereby identify an optimal composition group by traversing the entire design space (over 1,900,000 compositions). Our ML-model is validated by using 100 compositions in the high-activity region and 588 compositions in the low-activity region, which results in excellent recall values of nearly 100%. The predicted optimal amorphous high-entropy electrocatalyst demonstrates an ultralow overpotential of 159 mV at a current density of 10 mA cm for the alkaline OER in a 1 M KOH while exhibiting ultralong durability 10,218 h under a practical current density of 1 A cm in a 6 M KOH. Our work provides a general strategy for the automatic discovery and optimization of amorphous high-entropy oxyhydroxide electrocatalysts and could significantly impact the development of other amorphous high-entropy materials.