Machine learning-based prediction of supercapacitor performance for a novel electrode material: cerium oxynitride

Abstract From an engineering standpoint, specific capacity and cyclic stability may be considered the two most critical performance-related features for supercapacitor electrodes. The prediction of these two parameters is hence crucial for evaluating the prospect of a given material for a supercapacitor-electrode application. However, this prediction is highly non-trivial using existing atomistic approaches. As a solution, a combinatorial approach of value and grade prediction machine-learning models are used to predict the performance of a novel material (cerium oxynitride) for supercapacitor application. The model predicts the material to have a specific capacity of ~26.6 mAh g−1 and capacity retention of >90% for a particular material (morphology, composition, surface area) and operational (current density, applied potential window etc.) properties; which can be viably achieved via urea glass method. The experimental results (~26 mAh g−1 and ~100% capacity retention) considerably validate the predictive approach presented here. This article is the first instance wherein cerium oxynitride has been predicted and reported as a supercapacitor electrode. This makes the prediction and the validation made in this study of contemporary relevance.