Investigation for the influence of Sm2O3 and CeO2 nanoparticles on the microstructure and electrochemical behavior of epoxy and prediction of mechanical characterizations of adhesive joining of CFPEEK via machine learning

Abstract

With the increase demand for lightweight material combined with high mechanical and electrochemical properties, in this study we proposed a novel polymer nanocomposites (PNCs). Due to the unique characterizations of rare earth metal oxides nanoparticles, cerium oxide (CeO2)/samarium oxide (Sm2O3) were integrated to enhance microstructural, mechanical, electrochemical and the joint efficiency performance of epoxy resin matrix. The influences of various dispersion content of CeO2/Sm2O3 were explored. The adhesive joint efficiency and maximum shear force were determined by performing single lap joint through applying a thin layer of the synthesized PNCs to join carbon fiber polyetheretherketone (CFPEEK). Superior ultimate tensile strength was obtained by doping 7 wt% CeO2/Sm2O3 as the enhancement reached 483.482 % and 490.380 % respectively. The maximum joint efficiency reached 68.43 % and was achieved by doping 3 wt% of CeO2. An optimum enhancement in electrical conductivity was achieved by 1 wt% and 5 wt% of CeO2, while optimum enhancement in insulation or coating properties obtained by 5 wt% of Sm2O3. In addition, machine learning algorithms, including artificial neural networks, random forest, extreme gradient boosting (XGBoost), and k-nearest neighbors were applied to predict the investigated material properties. XGBoost provided robust predictions across both mechanical and electrochemical properties.