Dielectric Characterization and Machine Learning-Based Predictions in Polymer Composites with Mixed Nanoparticles

Abstract

Our research described in this manuscript investigated the dielectric properties and structural characteristics of Bisphenol-A epoxy resin composites infused with various concentrations (5 wt.%, 10 wt.%, and 15 wt.%) of hybrid nanofillers, namely alumina (Al₂O₃) and zinc oxide (ZnO). Ultrasonic dispersion was utilized to integrate the nanofillers into the resin matrix. Structural properties were assessed using X-ray diffraction (XRD), which confirmed the presence of the Al₂O₃ and ZnO nanoparticles within the epoxy matrix. Dielectric properties were measured over a frequency range of 10⁴ Hz to 2 MHz. The results provide new insights into the polarization mechanisms and structural characteristics of these composites, highlighting their potential for enhanced dielectric performance in high-frequency applications. To further understand and predict these dielectric properties, the CatBoost and LightGBM regression models were employed to predict the dielectric constant (ε'), loss tangent (tan δ), and AC conductivity (σ_ac) of these composites. The models demonstrated strong predictive accuracy, with performance metrics, including Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and R-squared (R²), indicating the robustness and accuracy of the models in predicting the dielectric properties of the composites. The study’s findings underscore the significant potential of Bisphenol-A epoxy resin composites with hybrid nanofillers for high-frequency dielectric applications.

Keywords:

• Hybrid nanofillers
• epoxy
• dielectric properties
• loss tangent
• machine learning
• CatBoost
• LightGBM

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The data supporting the outcomes of this study can be obtained directly from the corresponding author upon making a reasonable request.