Machine Learning Techniques to Analyze the Influence of Silica on the Physio-Chemical Properties of Aerogels

Abstract

This study explores the application of machine learning techniques, specifically principal component analysis (PCA), to analyze the influence of silica content on the physical and chemical properties of aerogels. Silica aerogels are renowned for their exceptional properties, including high porosity, large surface area, and low thermal conductivity, but their mechanical brittleness poses significant challenges. The study initially utilized cross-correlation analysis to examine the relationships between key properties such as the Brunauer–Emmett–Teller (BET) surface area, pore volume, density, and thermal conductivity. However, weak correlations prompted the application of PCA to uncover deeper insights into the data. The PCA results demonstrated that silica content has a significant impact on aerogel properties, with the first principal component (PC1) showing a strong positive correlation (R2 = 94%) with silica content. This suggests that higher silica levels correspond to lower thermal conductivity, porosity, and BET surface area, while increasing the density and elastic modulus. Additionally, the analysis identified the critical role of thermal conductivity in the second principal component (PC2), particularly in samples with moderate to high silica content. Overall, this study highlights the effectiveness of machine learning techniques like PCA in optimizing and understanding the complex inter-relationships among the physio-chemical properties of silica aerogels.

Keywords

• silica aerogels
• composite
• principal component analysis
• physical properties