Characteristics of nanoepoxy composite through structural, thermal and machine learning-enhanced dielectric analysis

This study aims to comprehensively explore the characteristics of hybrid nano fillers incorporated epoxy composites (NFIEC) using various analytical techniques. The research encompasses machine learning-enhanced dielectric analysis across a wide frequency spectrum ranging from 100 Hz to 2 MHz. The structural properties of the nano-fillers, pristine epoxy (PE), and NFIEC are investigated through X-ray Diffraction (XRD). Hybrid nano fillers, including combinations like SiO₂ + Fe₃O₄, ZnO + Fe₃O₄, TiO₂ + Fe₃O₄ and Al₂O₃ + Fe₃O₄, are added to the PE at a concentration of 1 wt.% to form NFIEC. Additionally, Differential Scanning Calorimetry are employed to gain further insights into the thermal behaviour of the NFIEC. This study also demonstrates the effectiveness of machine learning models, particularly Extra Trees, CatBoost, XGBoost, and Ensemble Averaging, in predicting dielectric properties (ε′ and ε′′) of NFIEC using data from LCR and VNA measurements. The models exhibited high predictive accuracy, with Ensemble Averaging consistently delivering robust performance across datasets, highlighting the potential of ML techniques in advancing material science by reducing experimental efforts and enhancing predictive precision.