Alkali-activated fly ash (AAFA) is being increasingly acknowledged as an eco-friendly binder free of Portland cement, valued for its low carbon footprint and promising engineering properties. However, the application of AAFA has been limited due to its unstable and uncontrollable engineering properties, which are closely tied to its microstructure. The microstructure of AAFA can be affected by several factors, including the intrinsic properties of fly ash, the types of alkaline activators, the mixture and curing regime. These factors can lead to various reactions, resulting in diverse microstructures, and thus a wide range of engineering properties. A deep understanding of the relationship between these influencing factors and the resulting microstructure is essential to bridge the gap between the mixtures and their corresponding engineering properties. Although the effects of these factors on the microstructure of AAFA have been extensively investigated experimentally, there is currently no numerical model capable of simulating the chemical reactions and microstructural development of AAFA. As a result, the reactions and microstructure, and thus the engineering properties of AAFA, remain unpredictable for a given mixture. Simulating the reactions and microstructural development would enable the customization of mix designs to achieve desired engineering properties, thereby promoting the application of AAFA. Therefore, the aim of this research is to simulate the reaction process and development of the microstructure of AAFA.
Thermodynamic modeling is a robust approach to simulate chemical reactions. However, the main challenge in thermodynamic modeling of AAFA lies in the lack of a thermodynamic database of its primary reaction product, N-(C-)A-S-H gel, which varies in Si/Al and Ca/Al ratios. Developing such a database requires accurate determination of the chemical compositions of N-(C-)A-S-H gels, which is difficult to achieve with conventional experimental techniques. Therefore, this research addresses this challenge by utilizing molecular dynamics simulations to determine the chemical compositions of N-(C-)A-S-H gels. By simulating the polymerization process that mimics actual reactions, the atomic structures of N-(C-)A-S-H gels with various Si/Al and Ca/Al ratios were constructed. According to the simulation results, it is proposed that N-(C-)A-S-H gels with a Si/Al ratio of 1-3 and a Ca/Al ratio of 0-0.5 can represent the chemical compositions of N-(C-)A-S-H gel in a mature AAFA paste.
After determining chemical compositions, synthesis of pure N-(C-)A-S-H gels is the second step to determine their thermodynamic data. However, synthesizing N-(C-)A-S-H gel with a Si/Al≥2 at a high pH (corresponding to the alkalinity range of pore solutions in AAFA paste), posed a double challenge. To address this issue, using a concentrated solution with an initial Si/Al ratio higher than the target is the key. Following this approach, N-(C-)A-S-H gels with a Si/Al ratio of 1-3 and a Ca/Al ratio of 0-0.5 were synthesized successfully and characterized by using XRF, XRD, FTIR, and TGA techniques. Subsequently, the solubility of the synthesized N-(C-)A-S-H gels was measured through a dissolution test. A thermodynamic database of N-(C-)A-S-H gels with various Si/Al and Ca/Al ratios was established for the first time, encompassing not only the solubility, but also the Gibbs free energy, heat capacity, entropy, enthalpy, and molar volume. This established thermodynamic database is the key to performing thermodynamic modeling to simulate the reactions of AAFA.
Coupled with the reaction kinetics determined by isothermal calorimetry and SEM-EDS analysis, the thermodynamic modeling of AAFA was performed for the first time to investigate the formation of reaction products and the phase assemblage of AAFA over time in GEMS software. The sodium hydroxide-activated system showed a close consistency between the modeling and experimental data regarding phase assemblage and pore solution chemistry, while for the sodium silicate-activated system, the simulated ion concentrations in the pore solution showed discrepancies compared to the experimental results. This discrepancy may be attributed to the high ionic strength in the sodium silicate-activated system, limitations in thermodynamic data of N-(C-)A-S-H gel and thermodynamic modeling approach itself.
To simulate the microstructure of AAFA, GeoMicro3D model, originally designed for alkali-activated slag, was extended to adapt to AAFA. To achieve this, first, the dissolution of fly ash in an alkaline solution was investigated experimentally, from which prediction functions were developed to describe the dissolution rate of Si and Al, accounting for the intrinsic characteristics of fly ash, solution pH, and temperature. The developed functions can accurately predict the dissolution behavior of fly ash, aligning well with the experimental results. Then, the GeoMicro3D model was extended by equipping with the thermodynamic database of N-(C-)A-S-H gels and the prediction functions for the dissolution of fly ash. GeoMicro3D was employed to simulate the reaction process and the 3D microstructural development over time of the sodium hydroxide-activated fly ash paste. The distribution of various phases in a 3D microstructure of AAFA can be captured and visualized over time. The simulated degree of reaction of fly ash and the porosity of AAFA were in good agreement with the corresponding experimental data. Furthermore, GeoMicro3D can well simulate the pore solution chemistry over time, consistent with the experimental results.
To sum up, the reaction and microstructure evolution of AAFA were investigated using multiple simulation techniques in this work. The extended GeoMicro3D model developed in this research paves the way for simulating the microstructure and the pore solution chemistry for any given AAFA mixture. This advancement contributes to a deeper understanding of the relationship between the AAFA mixture and the resulting microstructure. Furthermore, the mechanical properties, transport properties and durability of AAFA can be further evaluated based on the simulated microstructure constructed by using the extended GeoMicro3D. This model enables industries to effectively manage fly ashes of varying qualities and customize AAFA to meet specific engineering requirements. This not only improves the utilization of fly ash but also promotes the sustainability of construction practices.@en