Machine Learning assisted design of tailor-made nanocellulose films: A combination of experimental and computational studies

Nowadays, modern nanomaterial research is complemented by machine learning methods to reduce experimental costs and process time. With this motivation, here, we implemented artificial neural network (ANN), random forest (RF), and multiple linear regression (MLR) methods to predict the mechanical properties of three-component nanocomposite films consisting of polyvinyl alcohol (PVA) crosslinked 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) oxidized cellulose nanofibers (TOCNFs) and either ammonium zirconium carbonate (AZC) or glyoxal (Gx) using the mechanical properties of mono-component TOCNF films and two-component nanocomposites containing PVA, AZC, or Gx-crosslinked TOCNF as the input of prediction system. Prediction methods were evaluated with performance indicators and experimental data. Overall, MLR performed with least accuracy, whereas ANN prediction displayed the lowest error followed closely by RF. Additionally, the physically or/and chemically crosslinked hybrid films with optimized amount of crosslinkers resulted in structures with a strength to rupture that was significantly higher than that of the pure nanocellulose films (increases of up to ~90% in tensile strength and ~70% in Young's modulus). POLYM. COMPOS., 2019.