Machine Learning-Guided Analysis and Understanding of Nanocellulose for Polymer Composite Design

Abstract

Nanocellulose is a promising sustainable nanofiber material with outstanding properties such as light weight, high strength, and thermal stability. Its application as a reinforcing filler in polymer composites has attracted attention, but the diversity in morphology and complexity in composite processing make it challenging to achieve desired material properties. Herein, we demonstrate machine learning-guided approaches for analyzing nanocellulose structure and predicting composite performance. First, we present the development of machine learning models that predict the specific surface area and fiber morphology of nanocellulose from sedimentation data of aqueous dispersions. Furthermore, the impact strength of polypropylene composites was predicted based on the chemical and morphological features of different biomass-derived nanocelluloses. The proposed framework offers a powerful tool to accelerate composite material design and enables the effective utilization of diverse wood species and raw materials such as agricultural residues.