Abstract
The use of polymers in geotechnical engineering has accelerated rapidly due to their ability to enhance the mechanical and physical properties of soils. However, despite increasing interest, predicting the behavior of such materials across varying polymer concentrations remains a challenge and requires additional tools for an accurate evaluation. This study presents an integrated approach to the analysis and prediction of microstructural changes in polymer-modified soils that combines microscopy techniques with machine learning methods. A detailed microstructural analysis was conducted on soils modified with Xanthan Gum (XG) and Carboxymethyl Cellulose (CMC) using Scanning Electron Microscopy (SEM), Energy-Dispersive X-ray Spectroscopy (EDS), and elemental and morphological heatmap visualizations. These techniques allowed for a comprehensive investigation of microlevel changes occurring due to varying polymer concentrations. Based on experimental data, a linear regression model was developed to predict microstructural characteristics at a polymer concentration of 12%, an untested level in the laboratory. The results show that machine learning-based predictions derived from experimental data at lower concentrations (3%, 6%, and 9%) can effectively estimate microstructural parameters at higher concentrations. This approach offers a cost-effective and time-saving solution for the development of new and sustainable soil modification strategies. The optimized soil-to-polymer ratio is key to consistent microstructural modification.