Free-standing 3D Co3O4/CuCo-based Nanocomposite Electrodes for Boosting Hybrid Supercapacitors: A Combined Experimental and Machine Learning Insights

Developing a rational design of binder-free 3D nanocomposites is critical for substantially enhancing the performance of hybrid supercapacitors. In this study, Co₃O₄/CuCo-LDH, Co₃O₄/CuCo₂O₄, and Co₃O₄/CuCo₂S₄ hybrid nanocomposites were synthesized and directly grown on nickel foam via multiple hydrothermal processes. The synthesized hybrid electrodes exhibit abundant active sites and enhanced ion/electron diffusion, leading to superior electrochemical performance. The study is characterized by the fact that the metal impurities and metal used are constant to study the effect of functional groups on the composites. Particularly, Co₃O₄/CuCo-LDH and Co₃O₄/CuCo₂O₄ revealed excellent specific capacitances of 1558 F g⁻¹ (701 C g⁻¹ at 1 A g⁻¹) and 753.2 F g⁻¹ (376.6 C g⁻¹ at 1 A g⁻¹), respectively. Notably, Co₃O₄/CuCo₂S₄ achieved the highest specific capacitance of 2582.2 F g⁻¹ at 1 A g⁻¹. All of the fabricated materials combine and test in a full HSC device, and the Co₃O₄/CuCo₂S₄//CNF device achieves an impressive energy density of 84.05 Wh.kg-1, outperforming all other constructed devices at a power density of 390 W.kg-1. In artificial intelligence, the experimental galvanostatic charging-discharging (GCD) dataset is optimized to successfully predict capacitance and enhance energy optimization in AI-powered settings using numerous machine learning (ML) models. Six different ML models were applied: Least Absolute Shrinkage and Selection Operator (LASSO), k-nearest neighbors (KNN) regression, Extra Trees (ETs), Random Forests (RFs), and double models of deep neural networks (DNN). These models enable optimizing systems that will improve performance and maximize system utilization. The Shapley plot indicates that both time and current density can strongly affect each other positively and negatively, respectively. Nevertheless, it found that setting the maximum potential has a significant, though moderate, effect on prediction.