Fast material search of lithium ion conducting oxides using a recommender system†
Abstract
A practical material search using a recommender system is demonstrated to obtain novel lithium ion conducting oxides. The synthesis of unknown chemically relevant compositions (CRCs) proposed by the recommender system and their related materials effectively reveals two kinds of novel lithium ion conductors using different approaches. In the Li2O–GeO2–P2O5 system, Li6Ge2P4O17 is found, which has the same composition as the recommended unknown CRC. Less-than-10-time synthesis following the ranking order in the diagram provides evidence of the discovered new phase. In the other quasi-ternary diagram of the Li2O–ZnO–GeO2 system, Li3Zn0.65Ge4.35O10.85 is discovered by a combination of a recommender system and synthetic chemistry, because the composition of the novel phase is different from that of the recommended unknown CRC. Here too, the required time for material discovery is reduced to one-third of that required for random search without the recommender system. Phase identification and elemental analysis suggest that these discovered materials could have unique compositions and crystal structures. The room-temperature (∼300 K) ionic conductivity (10−9–10−6 S cm−1) of the novel phases can be improved by compositional and structural optimisations. The recommender system could emerge as a practical material search tool for enhancing the discovery rate of solid lithium ion conductors.
- This article is part of the themed collections: Editor’s Choice: Machine Learning for Materials Innovation and Journal of Materials Chemistry A HOT Papers