Data-Efficient Design of High-Entropy Oxygen Carriers for Chemical Looping Using Active Learning

TL;DR

This paper introduces a data-efficient method using active learning to design high-entropy oxygen carriers for chemical looping, speeding up material discovery.

Contribution

The novel contribution is an active learning strategy that combines predictive modeling and uncertainty estimation to explore complex material spaces efficiently.

Findings

The active learning approach accelerates discovery of high-entropy oxygen carriers more effectively than traditional methods.

The methodology is generalizable to other multicomponent material systems.

The strategy reduces time and data requirements for exploring compositional spaces.

Abstract

High-entropy materials,
first demonstrated in metallic alloys and
later extended to oxides and other systems, unlock a vast compositional
space with properties suited for catalysis, energy, and structural
materials. However, the high compositional complexity makes systematic
exploration challenging, and only a small portion of the design space
has been studied. To address this, we introduce an active learning
strategy that integrates predictive modeling, uncertainty estimation,
and iterative sampling to efficiently navigate embedded compositional
material spaces. This approach continuously learns from previous evaluations,
focusing subsequent searches on the most promising regions while reducing
both time and data requirements. We demonstrate this methodology in
the search for high-entropy oxygen carriers for chemical looping,
where it rapidly accelerates discovery and identifies promising candidates
more effectively than conventional trial-and-error or grid-search
approaches. Importantly, this strategy is general and well-suited
to exploring the vast space of multicomponent materials.

Figures