Electronic Origin of Doping-Induced Enhancements of Reactivity: Case Study of Tricalcium Silicate
Summary¶
Manipulating hydration reactivity of silicates in water remains challenging because surface initiation and bulk transport both matter. This study combines density functional theory with reactive molecular dynamics on triclinic T1 tricalcium silicate (C₃S), the most reactive Portland cement clinker phase. Stoichiometric and doped variants with Mg²⁺, Al³⁺, and Fe³⁺ substitutions are constructed to isolate dopant effects at fixed polymorph. DFT evaluates water sorption energetics on low-index surfaces, while ReaxFF MD extends trajectories to capture longer-term proton transport through the bulk. The abstract argues that initial hydration follows surface chemistry, whereas longer-term hydration is controlled by bulk proton transport, and that both stages correlate with dopant-induced electronic-structure changes.
Methods¶
Static QM (DFT). Bulk and slab models use VASP with PAW pseudopotentials and the PBE GGA functional (600 eV plane-wave cutoff, Γ-only k sampling, 0.02 eV/Å force tolerance, 10⁻⁶ eV energy tolerance as stated). T1–C\(_3\)S is used for both pure and doped crystals to isolate dopant chemistry from polymorph variability: Mg\(^{2+}\) substitutes on Ca\(^{2+}\) sites; Al\(^{3+}\) and Fe\(^{3+}\) are inserted pairwise on Ca and Si sites with overall charge neutrality; Fe cases use the preferred antiferromagnetic solution described in the article. (100) surfaces are cleaved preserving [SiO\(_4\)] tetrahedra; slabs use a 14 Å vacuum gap, and a two-unit-cell slab (2 × 162 atoms) is adopted after a thickness convergence check. Single water molecule adsorption is evaluated on symmetrically distinct surface sites with orientational variants (H-up / H-down) and results averaged to mitigate artificial dipole effects from the low-symmetry triclinic cleavage.
MD application (ReaxFF hydration). After DFT cleavage and relaxation, the paper drives reactive molecular dynamics (ReaxFF) on the same pure and doped T1–C\(_3\)S surface motifs. Bulk water is built by relaxing 405 H\(_2\)O molecules at ~0.99 g cm\(^{-3}\) in a triclinic box shaped to each crystal, first at 300 K under NVT-style thermostat control as stated. Those relaxed water cells are stacked along \(z\) onto the DFT surface cells (3D periodic supercells throughout). During hydration, a barostat drives the stress tensor normal component toward ~0 GPa along \(z\) while the in-plane lattice vectors stay fixed; production segments run 5 ns followed by 1 ns for statistics, all with Δt = 0.25 fs. The authors explicitly bound interpretation to nanosecond-scale hydration and omit slower processes such as dissolution/reprecipitation at microsecond horizons. MD engine (software name): N/A — the JPCC text extracted from the publisher PDF does not name an integration package (only ReaxFF as the reactive model). Replica exchange, applied electric fields: N/A — not used.
Force-field training. N/A — this paper applies an established ReaxFF parameterization for reactive Ca–Si–O–H chemistry (cited lineage in the article) rather than reporting a new fit.
Findings¶
Outcomes / mechanism. The abstract argues that early hydration tracks surface electronic/chemical features accessible to DFT, whereas longer-term hydration is dominated by proton transport into the bulk seen only after multi-nanosecond ReaxFF sampling—both stages correlated with dopant-induced electronic descriptors discussed in the paper.
Comparisons / levers. Mg / Al / Fe substitutions are used as controlled electronic-structure perturbations at fixed T1 structure; numerical adsorption trends, surface resonance correlations, and transport metrics are plotted in the JPCC article and should be quoted from there.
Limitations (authored). The manuscript notes PBE limitations, Γ-only sampling choices, and that nanosecond reactive MD still undershoots many curing-time phenomena (e.g., late-stage dissolution/reprecipitation) that may matter for engineering cements.
Limitations¶
DFT and ReaxFF each carry systematic errors; dopant space is limited to the constructed substitutions; nanosecond MD still undershoots laboratory hydration times.
Corpus notes¶
Cross-link this entry with [[2015hegoi-manzano-acs-am5b02505]] when reasoning about C₃S hydration: the Manzano study emphasizes dynamic disorder whereas Huang et al. emphasize dopant electronics—together they bracket composition and mechanism controls.
Downstream MAS prompts about “doped C₃S” should cite both papers when users ask for electronic-structure versus long-timescale reactive pathways, because each answers a different failure mode in simplified reactivity models.
Relevance to group¶
Template for paired QM + ReaxFF arguments on cement clinker phases.