Exploring the impact of dopants on ionic conductivity in solid‐state electrolytes: Unveiling insights using machine learning techniques

Abstract

Due to their high ionic conductivity, lithium lanthanum zirconium oxides (LLZO, Li7La3Zr2O12) of the garnet type are useful in a variety of applications and are good choice for solid state lithium‐ion batteries. The nature of dopants and their stoichiometry significantly impacts ionic conductivity. In this study, to explore the large design space of doped LLZO, we used optimized machine learning techniques based on random sampling screening of the Lazy classifier. Molecular, structural, and electronic descriptors were used to derive features for training the algorithms. The light gradient boosting machine and random forest algorithms exhibited a classification accuracy exceeding 95%. Notably, the relative density of LLZO was identified as the most correlated attribute to doped LLZO ionic conductivity. These findings highlight the potential of data‐driven algorithms in driving innovation and facilitating the development of novel materials.

Computational approach inspired advancements of solid-state electrolytes for lithium secondary batteries: from first-principles to machine learning

“…342 A similar approach was applied to the doped LLZO system, utilizing a training dataset of 208 samples and employing molecular, structural, and electronic descriptors as key features. 343 The relative density of LLZO was found to be the most critical factor influencing the ionic conductivity, followed by electronegativity. The HTS approach can also be combined with other numerical methods to create a multiscale platform for predicting SSE behaviors.…”