Fracture toughness of calcium–silicate–hydrate from molecular dynamics simulations

Mathieu Bauchy, Hadrien Laubie, Mohammad Javad Abdolhosseini Qomi, Christian G. Hoover, Franz‐Josef Ulm, R. J.-m. PellenqJournal of Non-Crystalline Solids 2015 · DOI · arXiv

Introduction

Cement is the most widely manufactured material on Earth, yet its production is responsible for approximately 7% of global CO2 emissions. To mitigate this environmental impact, there is a critical need to design "slender infrastructure"—structures that utilize less material while maintaining high durability and strength. Achieving this requires a fundamental understanding of the fracture toughness of cement's primary binding phase: calcium–silicate–hydrate (C–S–H).

While macroscopic data on cement exists, the intrinsic fracture properties of C–S–H grains at the nanoscale have remained elusive due to experimental measurement challenges. This study employs reactive molecular dynamics (MD) simulations to bridge this gap, providing a bottom-up foundation for designing tougher, more sustainable concrete.

Methodology

The researchers utilized a realistic, disordered model of C–S–H (the Pellenq-2009 model) with a stoichiometry of $(CaO)_{1.65}(SiO_2)(H_2O)_{1.73}$. To accurately simulate the breaking and forming of chemical bonds—essential for capturing the dissociation of water molecules during fracture—the REAXFF (Reactive Force Field) potential was employed.

Given the disordered nature and nanoscale ductility of C–S–H, standard Linear Elastic Fracture Mechanics (LEFM) formulas are often insufficient. Instead, the team applied the energetic theory of fracture, calculating the critical energy release rate ($G_c$) by integrating the stress-strain response during the full propagation of a crack under Mode I loading.

Stress-strain response during crack propagation in C-S-H, illustrating the mechanical behavior used to derive fracture toughness.

Nanoscale Ductility and Chemical Stabilization

A key finding of this research is that C–S–H is not perfectly brittle at the atomic scale. The molecular network exhibits significant deformation, creating a "plastic zone" (estimated at ~13.7 Å) around the crack tip that dissipates energy. Calculations indicate a brittleness parameter of 0.62; since a value of 1 represents perfect brittleness, this suggests that nearly 40% of the fracture energy is dissipated through plastic deformation.

Water plays a dual role in this process. While it depolymerizes silicate chains—making the structure easier to pull apart than pure silica—it also reacts with newly exposed, under-coordinated calcium and silicon atoms. This chemical reaction stabilizes the new surfaces and lowers the overall surface energy.

Potential energy evolution during the fracture process, highlighting the energy release (ΔU) and subsequent surface stabilization.

Key Results and Implications

The simulations yielded a fracture toughness ($K_{Ic}$) of $0.369 \pm 0.030 \text{ MPa}\cdot\text{m}^{1/2}$ and a critical energy release rate ($G_c$) of $1.72 \pm 0.29 \text{ J/m}^2$. Remarkably, this nanoscale $K_{Ic}$ value aligns closely with experimental measurements of macroscopic cement paste ($0.29–0.40 \text{ MPa}\cdot\text{m}^{1/2}$), identifying C–S–H grains as the dominant factor in cement's overall toughness.

Limitations and Future Outlook

While these results provide a robust nanoscale baseline, size effects remain a consideration. In much larger specimens, the material may transition to a "quasi-brittle" state where the plastic zone becomes negligible relative to the total volume. Future research will focus on upscaling these findings to meso-scale models and investigating how chemical tuning—such as adjusting the Ca/Si ratio—can further enhance the ductility and toughness of next-generation concrete.

Summary updated

BibTeX
@article{bauchy2015fracture,
  author = {Mathieu Bauchy and Hadrien Laubie and Mohammad Javad Abdolhosseini Qomi and Christian G. Hoover and Franz‐Josef Ulm and R. J.-M. Pellenq},
  title = {{Fracture toughness of calcium–silicate–hydrate from molecular dynamics simulations}},
  journal = {Journal of Non-Crystalline Solids},
  year = {2015},
  doi = {10.1016/j.jnoncrysol.2015.03.031},
}
Figure