Thermal behavior of coal fly ash geopolymers: structural analysis supported by molecular dynamics and machine learning methods

• Published in: Journal of thermal analysis and calorimetry Vol. 149; no. 10; pp. 4397 - 4409
• Main Authors: Król, M., Stoch, P., Szymczak, P., Mozgawa, W.
• Format: Journal Article
• Language: English
• Published: Cham Springer International Publishing 2024; Springer Nature B.V
• Subjects: Algorithms; Aluminosilicates; Aluminum silicates; Analytical Chemistry; Chemical composition; Chemistry; Chemistry and Materials Science; Cluster analysis; Compressive strength; Crystallization; Disintegration; Dynamic structural analysis; Fly ash; Geopolymers; Heat treatment; High temperature; In situ measurement; Infrared spectroscopy; Inorganic Chemistry; Machine learning; Measurement Science and Instrumentation; Microstructural analysis; Molecular dynamics; Nepheline; Physical Chemistry; Polymer Sciences; Sodalite; Sodium hydroxide; Sodium silicates; Thermodynamic properties; Thermal properties; Amorphous gel; Alkali-activation; Principal component analysis (PCA); Machine learning
• ISSN: 1388-6150; 1588-2926
• DOI: 10.1007/s10973-024-13004-y

This contribution presents the results of structural investigations on the cured and high temperatures of three series of geopolymers. The specimens were synthesized at 80 °C from coal fly ash and three activators of variable composition based on sodium hydroxide and sodium silicate solutions. Structural and microstructural analysis was performed, especially using in-situ measurements of XRD patterns and IR spectra as a function of temperature. The cured compounds' phase content and compressive strength changed slightly depending on the starting chemical composition. All analyzed materials experience a mass loss due to water removal at 300 °C, followed by increased porosity from disintegrating compounds above 300 °C. Higher alkali content improves strength (400–600 °C) possibly due to nepheline formation. The amorphous phase gradually softens during heating, influenced by alkali content. Structural analyses were supported by model calculations of a number of aluminosilicate structures. The cluster analysis using the k-means algorithm was used to divide the PCA space into regions that represent structural similarities, and analyzing specific point positions in the space allowed for several conclusions to be drawn about the studied materials, including that changes in chemical composition and thermal treatment can promote the transformation from sodalite to nepheline and that the glass network has elements that are nepheline-like and may promote its crystallization.