Advanced machine learning-driven characterization of new natural cellulosic Lablab purpureus fibers through PCA and K-means clustering techniques.
Journal:
International journal of biological macromolecules
PMID:
40037447
Abstract
The increasing demand for sustainable and eco-friendly materials has spurred significant interest in natural fibers as alternatives to synthetic reinforcements in composite applications. This study aims to explore the potential of Lablab purpureus fibers (LPFs) as sustainable materials by employing advanced characterization techniques and machine learning-driven analysis. Chemical analysis identified LPFs' primary composition as cellulose (72.34 %), hemicellulose (11.46 %), and lignin (8.99 %), with minor components including wax (3.45 %) and ash (2.59 %). The average fiber diameter was measured at 237.95 μm, with a density of 1.24 g/cm, making LPFs lightweight yet robust. Mechanical testing across varying relative humidity (RH) levels revealed a decrease in tensile properties, with fracture stress declining from 420 MPa at 24 % RH to 350 MPa at 81 % RH. X-ray diffraction (XRD) analysis demonstrated a crystallinity index (CI) of 74.62 % and a crystalline size of 8.73 nm, indicating high structural integrity. Fourier Transform Infrared (FTIR) spectroscopy, combined with Principal Component Analysis (PCA), provided insights into the chemical bonds within the fibers, confirming the presence of cellulose I and cellulose II polymorphs. Thermogravimetric Analysis (TGA) highlighted thermal degradation stages, with hemicellulose decomposition at 220-315 °C, cellulose decomposition at 315-400 °C, and lignin degradation above 400 °C, showcasing thermal stability up to 320 °C. Hydrothermal absorption behavior, analyzed through K-means clustering, revealed distinct absorption patterns, with a maximum moisture uptake of 12.3 % at 81 % RH. Biodegradability tests indicated increased decomposition with higher RH, peaking at 81 % RH with a weight loss of 68.57 % over 16 days. Scanning Electron Microscopy (SEM) revealed intricate fiber morphology, including layered transitions, internal voids, and a honeycomb-like surface structure. Compared to other natural fibers such as Cissus quadrangularis (CI: 82.73 %) and lavender (CI: 65 %), LPFs exhibit a balanced combination of mechanical strength, thermal stability, and biodegradability, making them promising candidates for biocomposites and eco-friendly materials. These findings, supported by machine learning-driven insights, position LPFs as a sustainable alternative to synthetic fibers in industrial applications.