The cement industry is a major contributor to global greenhouse gas emissions. Reducing these emissions is critical to meeting global climate goals. While improvements in energy efficiency have been pursued, a more radical approach involves modifying the cement material itself. Ordinary Portland cement's main constituents are alite and belite, impure forms of tricalcium silicate (C3S) and dicalcium silicate (C2S), respectively. The hydration of C3S and C2S is key to strength development, making understanding their hydration mechanisms crucial for designing environmentally friendly cements. Atomic and molecular-scale simulations offer a powerful tool to study the initial hydration process, excluding the influence of impurities. The initial step is considered to be the adsorption of water molecules onto the C3S and C2S surfaces, followed by proton diffusion and calcium ion detachment. However, the detailed dissolution pathway of calcium ions remains unclear, motivating this study which utilizes reactive force field molecular dynamics simulations to explore the initial hydration processes of C3S and C2S, aiming to reveal the general dissolution pathway of Ca ions.

Previous studies have investigated water molecule adsorption on C3S and C2S surfaces, indicating a preference for dissociative and molecular adsorption, respectively, in the case of single water molecule adsorption. However, for bulk water adsorption, molecular adsorption is preferred on C3S due to strong hydrogen bonding. C3S is generally more reactive than C2S, attributed to the more loosely bound ionic oxygen in C3S. After water adsorption, proton diffusion into the material and Ca ion detachment from the surface are key steps. However, the dissolution pathways of Ca ions, particularly the general pathway and the influence of multiple coupled chemical reactions, remained poorly understood; recent work only examined specific Ca ions in C3S.

This study employed reactive force field molecular dynamics (MD) simulations using the ReaxFF force field within the LAMMPS software. M3-C3S (010) and β-C2S (100) surfaces were selected for their prevalence in industrial clinkers and low surface energies. DFT calculations were used to optimize the unit cells and low-index surfaces before generating the simulation models. Large slab models were used to minimize size effects, with the vacuum region filled with water molecules. The systems underwent isothermal-isobaric (NPT) ensemble equilibration before continuous simulations at 298 K for up to 40 ns. A short time step of 0.1 fs was used to ensure accuracy. Umbrella sampling simulations were conducted to calculate the potential of mean force for the second dissolution process in C3S. Data analysis involved visualization using OVITO and calculations of radial distribution functions, atomic density, and pore size distributions. Python scripts were used for data analysis, and the zeo++ code was used for pore size distribution calculation.

The simulations revealed key differences in the initial hydration of C3S and C2S. For C3S, two distinct dissolution processes were identified for calcium ions. The first involved Ca-O bond breaking and formation of new Ca-Ow bonds (Ow represents oxygen from water). The second, slower process involved the detachment of these initially dissolved Ca ions from the surface, transforming through various coordination states (tridentate, bidentate, monodentate). Umbrella sampling simulations demonstrated that the hydroxylation of neighboring oxygen atoms is crucial for Ca ion desorption. A new "ligand teeth" structure was observed, where dissolved Ca ions are temporarily bound to oxygen atoms from water molecules adsorbed on other Ca ions. In contrast, for C2S, the simulations revealed primarily the first dissolution process within the 40 ns simulation timeframe. The solid-water interface was characterized for both materials, showing distinct layers (Guggenheim, Stern, diffuse, bulk) and the appearance of small pores (2-6 Å) in C3S facilitating proton transfer. The long-range order of C3S was found to be more significantly affected by hydration than C2S. The formation of ligand teeth structures was observed in both C3S and C2S, highlighting that the dissolution process isn't a simple bond-breaking/formation process but is influenced by the surrounding environment. The faster hydration of C3S compared to C2S in simulations aligns well with experimental data. The simulations also revealed hydroxylated species at the very beginning of hydration, which is consistent with experimental findings.

The findings address the research question by revealing the detailed, multi-step dissolution pathways of Ca ions in C3S and C2S, highlighting differences in reactivity between the two silicates at the molecular level. The discovery of the two-step dissolution process in C3S and the identification of the crucial role of hydroxylated surface oxygen atoms explain the observed faster hydration rate of C3S. The characterization of the solid/water interface reveals structural changes that may have significant implications for the overall cement hydration process and contribute to our understanding of the early hydration stages. The agreement between simulation results and experimental data validates the simulation methodology and provides confidence in the accuracy of the observed phenomena.

This study provides a comprehensive molecular-level understanding of the initial stages of cement hydration in C3S and C2S. Key contributions include the identification of a new Ca ion dissolution process and general dissolution pathways, the discovery of a new ligand teeth structure critical for Ca ion dissolution, and a detailed characterization of the solid/water interface. These findings advance our understanding of cement hydration and will serve as a valuable resource for the development of improved cement materials with reduced environmental impact. Future work could investigate other low-index surfaces of C2S and C3S and extend the simulation time to further elucidate the long-term hydration behaviors.

The simulations were limited to a 40 ns timeframe, which might not fully capture the complete hydration process. The models used idealized surfaces, and the effects of impurities and other cement components were not explicitly included. Further validation of the observed dissolution processes on other low-index surfaces and various Ca-based minerals is warranted.