Data-Driven Material Exploration of High Conductive Fluorides By Multi-Element Substitution

Abstract

Listen

Translate article

Barium fluoride (BaF2) has a fluorite-type structure and is attracting high attention as a solid electrolyte material for fluoride ion batteries that can be used under high voltage. Previous research has reported that the ionic conductivity increases by 4 to 5 orders of magnitude by partially replacing Ba in BaF2 with La. This indicates that elemental substitution is effective in improving the ionic conductivity of BaF2. However, single element substitution only explore much limited experimental space in elements, so that it should be difficult to discover innovative conductive compositions. By replacing various elements at the same time, the experimental space can be largely expanded and the possibility of finding the optimal composition increases. But, the combination numbers of multi-element substituted compositions are much over 10000, and exhaustive exploration of them is unrealistic. Therefore, we aimed to efficiently search for multi-element substitution compositions with high conductivity by creating a material-conductivity correlated exploring system under data-driven approach using large number of fluoride material database. The purpose of this research was to discover solid solution compositions in the multi-element substituted fluorides and understanding of the relationship between substitutional elements and solid solution formation.Approximately 20 types of multi-element substitution composition candidates were initially picked up using the D-optimal design (Dopt). The target compositions were synthesized to collect their XRD data. Next, we extracted the "number of peaks" from the XRD data as a feature values to express the degree of solid solution in the sample. Furthermore, we made a regression model with the element types and addition ratios as explanatory variables and the "number of peaks" as the objective variable. Using the regression model, approximately 20 candidate were predicted to form solid solution, in which we called as data-driven design of experiments (DOE) approach. Finally, the materials were actually prepared according to the proposal in DOE. The identification results were projected using the principal component analysis (PCA) to draw the material space.The obtained material space shows the number of peaks in the samples prepared using DOE decreased compared to those using Dopt. In detail, the Dopt-derived sample had 41.83 peaks, while the DOE-derived sample had 19.65 peaks on average. This result indicates that we were able to efficiently find preferable compositions that are advantageous for solid solution formation by using the data-driven method.In the presentation, we will also report the high-throughput composition exploration, their conductivities, and also development of virtual X-ray diffraction simulation system for over 10,000 kinds of multi-substituted composition of the fluorides. Acknowledgement s This research was partially supported by Knowledge Hub Aichi, the GIMRT Program of the Institute for Materials Research, Tohoku University (Proposal No. 202303-CRKKE-0048).