The melt‐mixing method in a twin screw extruder was utilized to develop blended nanocomposites of poly(lactic acid: PLA) and poly(butylene‐adipate‐co‐terephthalate: PBAT) with three combinations of weight percent 90/10, 70/30, and 50/50 (w/w) which contain 1, 3, and 5 phr of organic clays (Cloisite 30B). As a result, both x‐ray diffraction analysis and scanning electron microscopy revealed a consistent compatibilization effect of the nanoclays (NCs) within the PLA/PBAT matrix. The interfacial adhesion between the blend components is improved in matrix droplet structures, where the PBAT domains are evenly dispersed owing to the uniform dispersion of NCs at optimal 3 phr. Furthermore, NCs at this concentration acted as nucleating agents in the PLA/PBAT blends and enhanced the crystallization rate of the blend nanocomposites by lowering the cold crystallization temperature of the 90/10 blend by 5°C, as determined by differential scanning calorimetry (DSC). The neat PLA/PBAT blends displayed a drop in tensile strength and modulus, but their properties were restored when NCs were interfacially localized. In addition, elongation at the break point increased from ∼5.5% (70/30) to ∼5.9% (70/30/1). A comparison was also made between the theoretical values of the tensile modulus calculated using three‐phase mathematical models and the experimental results. According to the Carreau‐Yasuda model, the viscosity of blend nanocomposites increases with the addition of NC. The percolation threshold between the two filler contents was also elucidated as the storage modulus increased significantly from 70/30 (∼28 Pa) to 70/30/3 (∼2000 Pa) at low frequency. Furthermore, machine learning (ML) has been demonstrated to be an effective tool for data‐driven multi‐physical modeling that can be exploited to optimize rheological properties. The extreme gradient boosting (XGBoost) model yields the most accurate viscosity (η) prediction. Finally, the thermal stability of the blended components was examined via thermogravimetry analysis (TGA) and was found to be improved with the use of NCs.Highlights
Structure–property models were developed by experimental and computational data.
Young's modulus was predicted with high accuracy by Paul and Halpin‐Tsai models.
XGBoost modeled the rheological data well based on the machine learning technique.