Modeling and optimization of mechanical, water vapor permeability and haze properties of PLA and PBAT films reinforced with montmorillonite, halloysite nanotubes and palygorskite using artificial neural networks and genetic algorithms

Abstract

This study investigates the prediction and optimization of mechanical, barrier, and optical properties of biodegradable food packaging films by integrating artificial neural networks (ANN), multi-objective genetic algorithms (MOGA) and multicriteria decision making approaches. Poly (butylene adipate-co-terephthalate) (PBAT) and polylactic acid (PLA) polymers were reinforced with varying concentrations of montmorillonite (MMT), halloysite nanotubes (HNTs) and palygorskite to assess their effects on tensile strength, water vapor permeability, and haze, respectively representing mechanical, barrier and optical performance. MMT and HNTs were the most effective nanoclays in enhancing the mechanical, barrier, and optical properties of the films. The addition of HNTs decreased the water vapor permeability of PBAT by 43 % and increased its tensile strength up to 38.5 MPa at optimal concentrations. Similarly, incorporating MMT into PLA improved its water vapor barrier by 29 %. Experimental data were used to develop ANN models with strong predictive capabilities, achieving correlation factors exceeding 0.97. MOGA was then employed to generate a Pareto optimal solution front, illustrating the trade-offs between maximizing tensile strength and minimizing permeability and haze. The technique of order of preference by similarity to ideal solution (TOPSIS) method was applied to refine the selection of the optimal film. The analysis revealed that incorporating MMT into PLA at a concentration of 1.7 vol% optimally enhanced the film´s overall mechanical, barrier, and optical properties. This study highlights the potential of combining ANN and MOGA for optimizing biodegradable packaging films with balanced mechanical, optical and barrier properties.