Reactive force fields for aqueous and interfacial magnesium carbonate formation
Summary¶
Aqueous magnesium carbonate mineralization is a multistep problem coupling Mg²⁺ hydration, carbonate/bicarbonate speciation, and rearrangements at mineral–water interfaces. This Physical Chemistry Chemical Physics article develops Mg/C/O/H ReaxFF parameterizations in two coupled flavors: a bulk aqueous field that treats Mg²⁺ with fixed integer charge and uses nonbonded interactions plus Buckingham terms to ionic oxygen/hydrogen partners, and an interfacial field tailored for minerals and mineral–water contacts where surface chemistry dominates. The work targets scenarios relevant to CO₂ mineralization and low-temperature magnesite formation where interfacial water structure and proton transfer are first-order physics.
Methods¶
2 — Force-field training (Mg/C/O/H). The authors build ReaxFF in two coupled parts: a bulk aqueous set with Mg²⁺ on fixed integer charge and nonbonded + Buckingham O/H partners, and an interfacial set for minerals and mineral–water contacts. QM training data include brucite, magnesite, magnesia, MgH\(_2\), Mg\(_2\)C crystals, Mg²⁺–(H\(_2\)O)\(_n\) clusters, ion pairs, and water on mineral surfaces; see PCCP for DFT level, weights, and optimization workflow.
1 — MD application (ReaxFF in LAMMPS). Reactive MD of forsterite–(H\(_2\)O) and brucite–(H\(_2\)O) with HCO\(_3\)\(aq\))-related interfacial speciation, tracking H-transfer, HCO\(_3\)\(^-\)→CO\(_3\)\(^{2-}\)-like evolution, and interlayer OH\(...\)) dynamics (see abstract: ~0.22 eV “attachment” G-barrier language; exact definition in article). This wiki note is grounded in abstract-level reporting, so MD protocol slots are marked explicitly: Ensemble: N/A (NVE/NVT/NPT mode not stated in indexed text); Duration/stages: N/A (not stated); Barostat: N/A (not stated); Temperature schedule: N/A (not stated); Pressure control/target: N/A (not stated). Engine: LAMMPS (as reported). System focus: mineral-water interfaces noted above. Timestep, thermostat details, boundary conditions, and atom counts: N/A (not stated in indexed text). Electric / bias field and enhanced sampling (umbrella/metadynamics): N/A in abstract-level scope.
3 — Static DFT (standalone production work). N/A — the paper’s new DFT is embedded in the ReaxFF training/validation pipeline rather than a separate large static study summarized here.
4 — Experiments. N/A — CO\(_2\) mineralization discussion references literature experiments; the authors’ new evidence is ReaxFF-based.
Findings¶
Outcomes and mechanisms. At the forsterite–water interface with bicarbonate, ReaxFF MD shows long-range proton transfer that drives bicarbonate toward carbonate; the abstract reports a ~0.22 eV G-barrier for carbonate attachment to Mg sites, comparable in magnitude to aqueous Mg²⁺–CO\(_3\)\(^{2-}\) ion pairing in the same model. Hydroxide diffusivity in the interfacial water is anisotropic and heterogeneous (abstract). The authors use these results to discuss magnesite nucleation at low T in H\(_2\)O-poor, CO\(_2\))-rich interfacial conditions consistent with cited experiments. N/A in this note to list every water layer or all H-bond statistics — see PCCP figures/SI. Comparisons to literature and outlook are in the main text.
Limitations¶
Fixed Mg charge omits polarizable-ion physics that may matter in highly charged electric double layers or unusual solvation states; users should validate edge cases against higher-level electronic structure where feasible.
Reproducibility notes¶
Mineral nucleation runs should archive pH/CO₂ effective driving forces as implemented in the classical ensemble (even if implicit), water density near interfaces, and how carbonate species are initialized—proton-transfer cascades can depend on starting speciation. When reporting 0.22 eV barriers, cite the sampling method used for free energies (biasing vs umbrella) as stated in the article.
For forsterite surfaces, confirm that the modeled surface orientation and Mg/O termination match the experimental samples discussed—different cleavage planes expose distinct Mg densities and can change carbonate docking statistics even with the same force field. If comparing brucite versus forsterite interfacial water diffusivities, report water layer thickness and shear vs normal sampling windows separately—anisotropy can be a property of the analysis slab, not only the mineral.