Machine learning-assisted cross-domain prediction of ionic conductivity in sodium and lithium-based superionic conductors using facile descriptors

Yijie Xu,Yun Zong,Kedar Hippalgaonkar

doi:10.1088/2399-6528/ab92d8

Abstract

Solid state lithium- and sodium-ion batteries utilize solid ionically conducting compounds as electrolytes. However, the ionic conductivity of such materials tends to be lower than their liquid counterparts, necessitating research efforts into finding suitable alternatives. The process of electrolyte screening is often based on a mixture of domain expertise and trial-and-error, both of which are time and resource-intensive. In this work, we present a novel machine-learning based approach to predict the ionic conductivity of sodium and lithium-based SICON compounds. Using primarily theoretical elemental feature descriptors derivable from tabulated information on the unit cell and the atomic properties of the components of a target compound on a limited dataset of 70 NASICON-examples, we have designed a logistic regression-based model capable of distinguishing between poor and good superionic conductors with a validation accuracy of over 84%. Moreover, we demonstrate how such a system is capable of cross-domain classification on lithium-based examples with the same accuracy, despite being introduced to zero lithium-based compounds during training. Through a systematic permutation-based evaluation process, we reduced the number of considered features from 47 to 7, reduction of over 83%, while simultaneously improving model performance. The contributions of different electronic and structural features to overall ionic conductivity is also discussed, and contrasted with accepted theories in literature. Our results demonstrate the utility of such a facile tool in providing opportunities for initial screening of potential candidates as solid-state electrolytes through the use of existing data examples and simple tabulated or calculated features, reducing the time-to-market of such materials by helping to focus efforts on promising candidates. Given enough data utilizing suitable descriptors, high accurate cross-domain classifiers could be created for experimentalists, improving laboratory and computational efficiency.

Highlights

Lithium ion batteries (LIBs) have become the forefront of energy storage technology for a wide range of applications, including mobile devices and electric vehicles
As many of our models give the similare levels of validation and sodium test accuracies past n = 4, the selection criteria for the final optimal model feature were high accuracy, minimal overfitting on existing training data, and maximal generalization capability. The relationship between the former two factors can be controlled by comparing cross-validated accuracy (CVAC) with training accuracy (TAC), where a difference between training and validation accuracies would suggest a high level of overfitting
We have developed a simple, cost-effective screening algorithm for ionic conductivity for NASICON compounds using widely available, computationally simple elemental features

Summary

Introduction

Lithium ion batteries (LIBs) have become the forefront of energy storage technology for a wide range of applications, including mobile devices and electric vehicles. Na-ion batteries have shown great promise as being a safer, more cost-effective alternative for larger grid-storage applications. Both systems suffer from well publicized stability and safety issues [1, 2]. One of the these originates in their use of liquid electrolytes consisting of an ionic salt dissolved in an organic solvent, such as LiPF6 or NaPF6 in ethylene and dimethyl carbonate. While such electrolytes possess high ionic conductivity while remaining affordable, Category Examples. One of the primary issues in their development has been their lower ionic conductivities when compared to their liquid counterparts, often in the scale of several orders of magnitudes [4]

Objectives

Results

Discussion

Conclusion