Skip to content

Topological Control on the Structural Relaxation of Atomic Networks under Stress

Abstract

Accelerated MD on Pellenq C–S–H models with ReaxFF shows logarithmic shear creep under constant stress; relaxation propensity is minimized near isostatic networks without floppy modes or eigenstress—linking topology to aging.

Summary

Calcium–silicate–hydrate (C–S–H) controls creep, aging, and durability in cementitious materials, but linking atomistic disorder to macroscopic rheology requires models that capture both chemistry and long-time relaxation. Bauchy et al. construct Pellenq-type C–S–H structures across a range of Ca/Si ratios, introduce water via GCMC, and relax with ReaxFF, including reactive water–silicate interactions as cited in the letter. They then apply accelerated molecular dynamics under constant shear stress to extract creep kinetics and interpret trends with topological constraint theory (TCT) using floppy-mode counts, eigenstress, and isostaticity labels.

Methods

1 — MD application (atomistic dynamics)

Pellenq-type calcium–silicate–hydrate (C–S–H) models spanning Ca/Si from ~1.0 to 1.9 are prepared by defecting an 11 Å tobermorite parent and grand canonical Monte Carlo (GCMC) addition of interlayer water at room temperature and constant volume, followed by ReaxFF relaxation that allows water dissociation to hydroxyl species (methodology pointer to Pellenq et al. as cited in the letter). Prepared samples are stress-relaxed at 300 K before creep loading. Production creep uses ReaxFF molecular dynamics with an integration timestep of 0.25 fs (stated in the letter text) and accelerated molecular dynamics to reach laboratory-relevant relaxation timescales under constant shear stress (acceleration variant and run lengths detailed in the letter + Supplemental Material).

  • Engine / code: ReaxFF reactive molecular dynamics as cited in the letter (specific MD engine not named on the indexed pages).
  • System size & composition: Layered C–S–H nanograin models with tunable Ca/Si and interlayer H₂O/OH⁻ content as generated from the tobermorite defect + GCMC hydration protocol summarized above.
  • Boundaries / periodicity: 3D periodic supercells consistent with the Pellenq C–S–H construction workflow (see cited Refs. [27–28] in the letter).
  • Ensemble / stress: Initial zero-stress relaxation at 300 K; subsequent constant shear stress creep protocol using accelerated MD (details in SI).
  • Timestep: 0.25 fs (explicitly stated in the indexed letter excerpt).
  • Duration / stages: N/A on this wiki page — total accelerated-MD clock and equivalent real time are not reproduced here; use the PRL SI bundle.
  • Thermostat: N/A — explicit thermostat name not in the indexed letter excerpt (likely embedded in the accelerated-MD prescription in the SI).
  • Barostat: N/A for the quoted creep segment — shear stress control dominates; hydrostatic NPT usage, if any, is not restated on the indexed pages.
  • Temperature: 300 K for the systematic pre-creep relaxation noted in the letter.
  • Pressure: N/A — creep narrative emphasizes shear stress; isotropic pressure targets are not summarized on the indexed excerpt.
  • Electric field: N/A — not used.
  • Replica / enhanced sampling: Accelerated molecular dynamics (hyperdynamics / parallel-replica class—exact flavor in PRL/SI) replaces brute-force integration for rare activated events.

2 — Force-field training

N/A as a new fit in this letter — the study consumes an established ReaxFF parametrization for C–S–H with reactive water chemistry (cited parameter lineage).

3 — Static QM / DFT-only

N/A — DFT is not the long-timescale creep engine; QM appears only via prior validation literature referenced for the parent models.

4 — Topology analysis (non-MD)

Topological constraint theory (TCT) metrics—floppy-mode counts, eigenstress, and isostaticity labels—are computed on the same atomic networks to interpret shear creep propensity.

Findings

Outcomes and mechanisms

Under constant shear stress, C–S–H exhibits delayed logarithmic shear creep analogous to glassy network relaxation. Isostatic compositions—simultaneously lacking soft floppy modes and large built-in eigenstress—show the lowest propensity for stress-driven relaxation, whereas flexible (n_c < 3) and stressed-rigid (n_c > 3) networks creep faster via soft modes or internal stress release, respectively.

Comparisons

The letter contrasts these TCT states using the Pellenq C–S–H model family, which the cited prior work validates against nanoindentation modulus/hardness experiments—supporting the use of the same models for creep trends.

Sensitivity / design levers

Ca/Si ratio (and the associated defect + hydration history) shifts the network between flexible, isostatic, and stressed-rigid regimes, which the authors map directly to creep propensity.

Limitations, outlook, and corpus honesty

PRL brevity means full accelerated-MD parameters, statistical replication, and additional sensitivity sweeps live in the Supplemental Material and referenced longer papers—this wiki entry is not a substitute for those tables. ReaxFF omits explicit electronic-structure detail; the authors flag that limitation when extrapolating to observables dominated by charge transfer or redox chemistry.

Limitations

PRL letter format—full simulation parameters in supplementary references; single potential landscape (ReaxFF) governs chemistry. Readers seeking quantitative creep rates should treat this page as an orientation summary and extract numbers from the letter + SI bundle referenced by the DOI. C–S–H nanostructure varies with curing history; the modeled Ca/Si sweeps should not be read as exhaustive coverage of cement chemistry space.

Relevance to group

Uses ReaxFF cement parametrizations and TCT tools aligned with C–S–H modeling in the broader KB.

Citations and evidence anchors

  • DOI: 10.1103/PhysRevLett.119.035502