A Generalized Machine Learning Model for Predicting Ionic Conductivity for Ionic Liquids

21 March 2022, Version 1
This content is a preprint and has not undergone peer review at the time of posting.


Ionic liquids are currently being considered as potential electrolyte candidates for next-generation batteries and energy storage devices due to their high thermal and chemical stability. However, high viscosity and low conductivity at lower temperatures have severely hampered their commercial applications. To overcome these challenges, it is necessary to develop structure-property models for ionic liquid transport properties to guide the ionic liquid design. This work expands our previous effort in developing a machine learning model on imidazolium-based ionic liquids to now include ten different cation families, representing structural and chemical diversity. The model dataset contains 2869 ionic conductivity values over a temperature range of 238-472 K collected from the NIST ILThermo database and literature values for 397 unique ionic liquids. The database covers 214 unique cations and 68 unique anions. Three machine learning models, multiple linear regression, random forest, and extreme gradient boosting, are applied to correlate the ionic liquid conductivity data with cation and anion features. Shapely additive analysis is performed to glean insights into cation and anion features with significant impact on ionic conductivity. Finally, the extreme gradient boosting model is used to predict ionic conductivity of ionic liquids from all the possible combinations of unique cations and anions to identify ionic liquids crossing the ionic conductivity threshold of 2.0 S/m


Ionic conductivity
ionic liquids
machine learning

Supplementary materials

Supporting information
Contains machine learning model hyper parameter tuning, performance of the models for training and test data sets, external data set prediction and Shapely additive feature analysis


Comments are not moderated before they are posted, but they can be removed by the site moderators if they are found to be in contravention of our Commenting Policy [opens in a new tab] - please read this policy before you post. Comments should be used for scholarly discussion of the content in question. You can find more information about how to use the commenting feature here [opens in a new tab] .
This site is protected by reCAPTCHA and the Google Privacy Policy [opens in a new tab] and Terms of Service [opens in a new tab] apply.