A complete and effective target-based data-driven flow screening for reliable cathode materials for aluminum-ion batteries

Rechargeable multivalent aluminum-ion batteries (AIBs) are expected to be the alternative energy storage batteries with great promise for future development due to the abundance of aluminum elements and low cost. However, the current lack of high voltage, high capacity, high transportability, and high energy density AIBs cathode materials is a major impediment to the practical development. In this work, we develop a comprehensive and effective data-driven workflow with the starting point of screening for more reliable cathode materials for multivalent AIBs. The proposed workflow is mainly supported by machine learning (ML) algorithms and deep learning framework. Driven by data from density functional theory (DFT) calculations, and additional experimental data from the literature are added to correct for the workflow’s model errors. In the context of the current poor availability of data on various properties of AIBs, a database of 1470 promising novel inorganic cathode materials for AIBs has been created. It provides the selected material’s performance in terms of voltage, transportability. A flexible framework for extending other important unexplored features is also developed, including theoretical specific capacity (&C), energy density (&E), and max volume change parameters (&M). Finally, based on the excellent experimental performance of the -based, portions of which are subjected for DFT calculation for verifying the workflow’s interpretability, and all of the selected -based obtain a better voltage plateau with a lower diffusion barrier. The presented work demonstrates a valuable experimental reference for the progress of cathode materials for AIBs and offers possible new avenues for accelerating the progress of inorganic cathode materials.