Development of microstructure-rheological and electrical properties relationship in PS/POE/HNTs blend nanocomposites using machine learning

• Published in: Polymer testing 2024-08, Vol.137, p.108503, Article 108503
• Main Authors: Estaji, Sara, Akbari, Homa, Tayouri, Mohammad Iman, Miri, Fatemeh Sadat, Salahshoori, Iman, Ruckdaschel, Holger, Dawi, Elmuez A., Khonakdar, Hossein Ali
• Format: Article
• Language: English
• Subjects: Electrical properties; Halloysite nanotubes; Polyolefin elastomer; Polystyrene; Rheological properties
• ISSN: 0142-9418; 1873-2348
• DOI: 10.1016/j.polymertesting.2024.108503

Halloysite nanotubes (HNTs) and polypropylene-grafted maleic anhydride (PP-g-MA) were studied for their effects in blends of polystyrene (PS) and polyolefin elastomer (POE). The method used to prepare PS/POE blends (90/10 and 80/20 wt/wt) containing 1, 3, and 5 phr HNTs with or without PP-g-MA (a compatibilizer) was melt blending. Structural and morphological studies using X-ray diffraction analysis (XRD), scanning electron microscopy assisted energy dispersive X-ray spectroscopy (SEM-EDS), and transmission electron microscopy (TEM) confirmed a matrix-droplet morphology and the sample containing compatibilizer has better microstructure than the other formulations. The presence of both PP-g-MA and HNTs together has been discovered to improve the viscoelastic properties of the solid, as evidenced by increased storage modulus and complex viscosity. A notable change occurred in the rheological behavior of the PS/POE blends containing 5 phr of PP-g-MA and HNTs. The dependence of zero-shear viscosity on HNTs loading (0–5 phr) was approximated as a polynomial curve by fitting the experimental data with the Carreau-Yasuda model. Computational fluid dynamics (CFD) simulations were also used to study the effects of HNTs loading on changes in fluid flow patterns and shear rates. The calculated effective viscosities at a given shear rate (0.05 1/s) were in qualitative agreement with the experimental results. Moreover, we utilized various machine-learning techniques to predict the complex viscosity of PS/POE blends and their nanocomposites. The results showed that Extreme Gradient Boosting (XGBoost) outperformed other predictive models based on evaluation metrics. Four-point probe measurements found that the samples containing 5 phr HNT had the lowest conductivities due to the presence of aggregated structures. However, the homogeneous distribution of HNT led to a sudden rise in conductivity in the samples containing 5 phr of PP-g-MA. Computer modeling results of samples with uniform and non-uniform HNT distributions showed that the conductivity decreased with HNT loading and decreased considerably compared to the uniform distribution. •A notable change took place in the rheological behavior of the PS/POE blends containing 5 phr of PP-g-MA and HNTs.•The calculated effective viscosities at a given shear rate (0.05 1/s) were in qualitative agreement with the experimental results.•Utilized various machine-learning techniques to predict the complex viscosity of PS/POE blen