High-throughput atomistic modeling of nanocrystalline structure and mechanics of calcium aluminate silicate hydrate

Yunjian Li, Cheng Chen, Zhenning Li, Zongjin LiNature Communications 2025 · DOI

TL;DR

This paper introduces CASHgen, a novel high-throughput atomistic modeling framework to simulate calcium aluminate silicate hydrate (C-A-S-H), a key component in environmentally friendly cements. By systematically exploring a wide range of compositions, the study reveals how Ca/Si and Al/Si ratios dictate C-A-S-H structure and mechanical properties, offering crucial insights for designing stronger, low-carbon cementitious materials.

What Problem Does This Paper Address?

Cement production is a major contributor to global CO2 emissions. While aluminum-containing cements offer a more sustainable alternative, the fundamental understanding of their primary hydration product, calcium aluminate silicate hydrate (C-A-S-H), at the atomic level remains limited. This lack of understanding hinders the optimization of C-A-S-H's structure and mechanical properties, which are crucial for developing high-performance, low-carbon cementitious materials. Existing models struggle with the complex chemical composition and inherent disorder of C-A-S-H, making it difficult to predict its behavior across various compositions.

Methods

The core of this research is the development of CASHgen, an in-house automatic structure generation program. This program leverages a "building block" approach, similar to previous work on C-S-H, but adapted to incorporate aluminum. CASHgen systematically constructs a vast number of C-A-S-H atomic structures by combining predefined "genes" (atomic groups) and allowing for variations in Ca/Si and Al/Si ratios, as well as defect structures.

Following structure generation, molecular dynamics (MD) simulations were employed to investigate the structural and mechanical properties of these generated C-A-S-H models. The study used the ericaFF2 force field for these simulations. Mechanical properties were analyzed using the stress-strain method and the Voigt-Reuss-Hill (VRH) method for calculating elastic moduli. Pair distribution functions (PDFs) and radial distribution functions (RDFs) were also computed to analyze atomic ordering.

Datasets

The study generated and analyzed 1600 distinct C-A-S-H atomic structures. These structures were systematically created to cover a broad range of Ca/Si ratios (1.3-1.9) and Al/Si ratios (0-0.15). Additionally, a subset of structures with interlayer aluminum (Al(V)) was specifically generated to reflect experimental observations. The generated structural data, along with data for mechanical properties analysis, are made available on figshare.

Key Insights and Ideas

  • CASHgen as a powerful tool: The development of CASHgen represents a significant step forward, enabling the systematic and automated generation of complex C-A-S-H structures with controlled compositions and defects. This overcomes previous limitations in atomic-level modeling of C-A-S-H.
  • Composition dictates structure and mechanics: The study clearly demonstrates that both the Ca/Si and Al/Si ratios are critical determinants of C-A-S-H's mean chain length (MCL), interlayer spacing, and ultimately, its mechanical performance.
  • Optimal Ca/Si ratio for performance: C-A-S-H exhibits its best mechanical properties around a Ca/Si ratio of approximately 1.5. Deviating from this ratio, particularly increasing Ca/Si, leads to increased disorder and reduced stiffness.
  • Aluminum's beneficial role: Increasing the Al/Si ratio has a positive impact on mechanical performance by promoting longer silicate chains (higher MCL) and enhancing stiffness. This highlights aluminum's crucial role in strengthening C-A-S-H.
  • Interlayer dynamics are key: The study reveals a complex interplay of calcium and aluminum within the interlayer region, significantly influencing interlayer spacing and contributing to the anisotropic mechanical behavior of C-A-S-H.

Key Results

  • Structural Characterization: The study accurately captured key C-A-S-H characteristics like MCL, 2H/Si, and Ca-OH/Ca ratios, showing good agreement with experimental data.
  • Mechanical Performance:
    • Optimal mechanical performance was observed at a Ca/Si ratio of ~1.5.
  • Increasing Al/Si ratio generally led to improved mechanical properties due to increased chain polymerization.
  • Anisotropy in mechanical properties was clearly demonstrated, with different responses along the X, Y, and Z directions depending on composition.
  • Interlayer Spacing: Interlayer spacing generally increased with higher Ca/Si and Al/Si ratios, with a complex relationship influenced by the presence of interlayer calcium and aluminum.
  • Density: Density initially increased from Ca/Si = 1.3 to 1.5 and then decreased, attributed to the balance between structural compactness and the presence of interlayer water and ions.

Limitations and Future Work

  • Force Field Limitations: While the ericaFF2 force field was used, the study acknowledges that classical MD simulations have inherent limitations. The observed differences in interlayer distance before and after equilibration suggest that condensation processes might be insufficient, and more advanced force fields or machine learning potentials derived from DFT could offer greater precision.
  • Experimental Data Gaps: The study notes that experimental data for C-A-S-H, particularly for low Ca and high Al conditions, is relatively scarce, which can limit direct validation in certain compositional ranges.
  • Future Work: The authors suggest that the CASHgen program can be extended to model other nanocrystalline materials with atomic substitutions or dopants. The developed framework and generated data can serve as a foundation for creating a "genetic database" for C-A-S-H nanomaterials, enabling bottom-up optimization of their performance for advanced, low-carbon cementitious applications.

Summary updated

Abstract

Although aluminum-containing cements have gained attention as environmentally friendly construction materials, the nanocrystalline structure and mechanical behavior of their primary hydration product, calcium aluminate silicate hydrate (C-A-S-H), remain poorly understood due to its complex chemical composition and structural disorder. Here, we present a high-throughput atomistic modeling framework to systematically investigate the structural and mechanical properties of C-A-S-H across a broad range of Ca/Si (1.3-1.9) and Al/Si (0-0.15) ratios. The compositional, structural, and mechanical features of C-A-S-H are accurately captured by molecular dynamics simulations of 1600 distinct C-A-S-H structures constructed using our in-house automatic structure generation program, CASHgen. Our findings highlight the influence of Ca/Si and Al/Si ratios on key C-A-S-H characteristics, including the mean chain length (MCL), interlayer spacing, coordination number and elastic moduli. Specifically, C-A-S-H exhibits optimal mechanical performance at a Ca/Si ratio of approximately 1.5, while further increases in Ca/Si introduce disorder and reduce stiffness. In contrast, increasing the Al/Si ratio promotes chain polymerization, leading to longer MCLs and improved mechanical performance. These results provide atomic-scale insights into the structure-property relationships in C-A-S-H and offer design guidelines for high-performance, low-carbon cementitious materials.
BibTeX
@inproceedings{li2025high,
  author = {Yunjian Li and Cheng Chen and Zhenning Li and Zongjin Li},
  title = {{High-throughput atomistic modeling of nanocrystalline structure and mechanics of calcium aluminate silicate hydrate}},
  booktitle = {Nature Communications},
  year = {2025},
  doi = {10.1038/s41467-025-60631-6},
}
Figure