Machine Learning Potential for Calcium Silicate Hydrates With Broad Compositional and Structural Diversity

Yunjian Li, Cheng Chen, Roland J‐M Pellenq, Zongjin Li, Jiaping Liu · 2026

Read Machine Learning Potential for Calcium Silicate Hydrates With Broad Compositional and Structural Diversity by Yunjian Li, Cheng Chen, Roland J‐M Pellenq, Zongjin Li, Jiaping Liu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/PMC13618771.txt.

Description

Yunjian Li and four colleagues build a machine learning potential that lets computers simulate cement's binding glue, calcium silicate hydrate, with near-quantum accuracy. This disordered nanoscale gel decides concrete's strength and durability, yet its shifting chemistry defeats simple models. The team trained their potential on more than fifty thousand quantum-labeled cases drawn from nearly seven hundred distinct mixes spanning defects and hydration states.

The resulting model reproduced structural, dynamic, and mechanical properties close to full quantum calculations. It settled long-standing disagreements with high-pressure X-ray experiments, and transferred reliably to large, disordered, and nanoporous systems it had never seen. It also runs much faster than the standard reactive force field it replaces. The authors frame it as a foundation for cement modeling at many scales, a data-driven cement genome, and lower-carbon construction materials.

Catalog record

CREATORYunjian Li, Cheng Chen, Roland J‐M Pellenq, Zongjin Li, Jiaping Liu
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE57 KB
LICENSEhttps://creativecommons.org/licenses/by/4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/article/MED/PMC13618771

Browse the full library · About the Tower of Babel