Energy storage is required in scale to enable the transition from the current carbon-based economy to one based on renewable energy. Hence shifting towards a paradigm wherein earth-abundant materials are used for developing energy storage technologies is essential. Ceria is among the rare earth oxides that are well distributed globally, and it exhibits good redox behavior (Ce4+/Ce3+). However, ceria also suffers from a bottleneck. It provides only moderate specific capacitance; hence given its relatively abundant global distribution, it deserves attention, especially for improving its performance in supercapacitor applications.

Given the above, this thesis focuses on three major themes with ceria as the selected material: (i) capacitive enhancement, (ii) computational performance prediction driven materials design, (iii) data-driven prediction of material performance.

Initially, we demonstrate two approaches to enhance the supercapacitive performance of ceria: (a) the role of doping and (b) the use of redox additives in the traditional alkaline aqueous electrolyte. A dopant (with primary oxidation states as 4+ and 3+) is found to aid in the enhancement of (A) the oxygen-vacancy concentration and (B) lattice oxygen mobility, compared to pristine ceria. Hence, doping standalone is shown to improve specific capacitance over pristine ceria significantly.

A Mott-Littleton-based method is then utilized to choose the more suitable dopant. The dopant design is then validated in a practical device. A computational method is used for the prediction of the dopant. The result is thereafter verified experimentally. Followed by the verification, the prediction is further validated through an electrochemical study. Consistent with computation, Cr-doped ceria is found to be a champion compared to other materials in this study (Eu-doped ceria and pristine ceria). Finally, the Cr-doped ceria provided ~100% specific capacitance retention after 10,000 cycles with moderate leakage current (~17 μA) and high-rate capability (~83% of capacitance retention even after increasing the current density by a factor of eight).
After that, we demonstrate redox-additive-based [K3Fe(CN)6] enhancement of charge-storage capacity. However, despite the enhancement (also consistent with recent reports from other groups), from a technology-development standpoint, we show that there are challenges associated with a lack of standard measurement techniques to evaluate device parameters associated with redox-additives. To address this concern, we have developed a new method that delineates the contribution of the effective mass of the redox additive on the electrode surface. This method is expected to be generic and will be used to evaluate other redox additive-based supercapacitor devices.

Then, a machine learning-based paradigm is used to predict the material’s (cerium oxynitride) specific capacitance and cyclic stability. Significant outcomes of our proof-of-concept ‘prediction validation’ work on ceria-based supercapacitors are, (I) a single parameter – namely the oxygen vacancy formation energy - as elucidated by the Mott-Littleton method is found to be a good measure of suitability, and (II) data-driven approach is proven to be handy in predicting the performance of a novel material for supercapacitor application. The thesis ends with experimental validation of the performance as predicted by machine learning.