Atomistic simulations for the prediction of physical properties of cement-based materials.

摘要

A number of different complex reactions occur during the hydration of cement i.e., when an anhydrous cement powder is mixed with water. The hydration process yields a porous, complex multi-phase, heterogeneous and amorphous cement paste matrix. One of the most important hydration products of the cement-based materials is Calcium silicate hydrates (C-S-H) which constitutes about 60-70% of completely hydrated cement. Hence, C-S-H is believed to be responsible for the properties (such as hardness, cohesion, strength etc.) of the cementitious materials. If the individual atoms in the structure of C-S-H could be manipulated at the nano scale, it is possible to obtain better control over macroscopic material properties of the cement-based materials. Ultimately, this could result in new products which are more durable and stronger. Unfortunately, due to the amorphous nature of C-S-H, its atomic structure is not well understood. A comprehensive understanding of its chemistry (atomic structure) could lead to an improved control of their material properties at the nano level. But this is not an easy task because the continuous changes that occur in the structure of C-S-H with respect to time and environment make experimental examination of its structure extremely difficult. Even with the use of sophisticated experimental techniques such as Powder Diffraction (X-Ray, Electron and Neutron Diffractions), Nuclear Magnetic Resonance, Scanning Electron Microscopy, Atomic Force Microscopy, the structure of C-S-H is not completely understood. However, some poorly defined, yet, consistent patterns are already identified, that C-S-H is a layered structure comprising Calcium and water molecules sandwiched between the tetrahedral silicate chains. Hence, a computational approach is sought in this work to investigate the atomic structure of C-S-H which in turn will help in understanding the related experimental studies.;In this research work, as a start, the atomic structure of C-S-H is modeled with the help of atomic configurations of some naturally existing perfect crystals like Tobermorite, Jennite etc. that have closer structural similarities with C-S-H. The mechanical properties (such as Young's modulus, Poisson's ratio, Bulk modulus, and Shear modulus) of these structures have been determined with the help of Energy Optimization studies through GULP code. The calculated bulk properties of the crystalline C-S-H models are found to be 4 to 5 times greater than the existing experimental data for the bulk cement pastes. The reasons for these exaggerated values have been investigated and analyzed. Some probable explanations for the over estimation could be: (a) Defect free C-S-H model considered for the study; (b) Inability to account porosity in the model.;Two different methods are presented in this work to resolve the aforesaid problems and to make the results more realistic and comparable with the experimental data. Firstly, the atomic structures of C-S-H are modeled with finite length of silicate chains (dimmer) as it is not a perfect crystal. Secondly, a packing factor is included in the calculation of bulk properties using Mori-Tanaka equations inorder to consider the porous nature of the cement-based materials. After making these two amendments, the calculated bulk properties of C-S-H are then in close agreement with the experimental data. Also, a possible predictive structure for C-S-H is also proposed. Our computations show that the length of the silicate chains has a significant effect on the bulk properties of C-S-H.