Predicting the salt adsorption capacity of different capacitive deionization electrodes using random forest

Abstract

Capacitive deionization (CDI) is an emerging technology in brackish water desalination. Several types of CDI electrodes have been fabricated and tested throughout the years and most, if not all, require hours of CDI performance testing to assess their capability. In this study, a model was created to predict the salt adsorption capacity of different carbon-based CDI electrodes. Random forest, a statistical classification method, was utilized in the making of the model. Based on the results, the generated models had high accuracy and precision in estimating the salt adsorption capacity, representing R2 values mostly over 0.9. Furthermore, feature importance analysis showed that specific capacitance at higher scan rates can reduce the performance of the model. Therefore, it is recommended that low scan rate specific capacitance be used in creating any kind of CDI model. This study also illustrated that a universal model is possible for CDI with a minor sacrifice in precision. Depending on the goal of the end user, individual and universal models can be used in the prediction of salt adsorption capacity of various CDI electrodes.