Mixing I and Br in inorganic perovskites: atomistic insights from reactive molecular dynamics simulations
Corpus note
The repository holds a galley PDF; cite the journal version of record via the DOI.
Summary¶
The paper extends a CsPbI₃-focused ReaxFF parameterization to CsPbBr₃ and mixed halide CsPb(BrₓI₁₋ₓ)₃, then applies large-scale molecular dynamics (MD) to phase transitions and octahedral dynamics in all-inorganic lead halide perovskites. Co-authors include Adri C. T. van Duin with Dutch partners Calero and Tao. The motivation connects mixed halide compositions to bandgap tuning and stability questions in photovoltaics, where anharmonic lattice dynamics and halide heterogeneity complicate purely harmonic phonon pictures—motivating explicit ReaxFF sampling of octahedral motion (introduction themes; abstract).
Methods¶
ReaxFF parameterization (A)¶
DFT training (VASP, ADF, PBE + D3(BJ)): charges, EOS (perovskite + non-perovskite), precursor EOS (CsX, PbX\(_2\)), defect E\(_f\)/barriers, phase-transition targets.
Large-scale reactive MD (B)¶
LAMMPS ReaxFF on large cells/long runs for T-driven cubicization, octahedral motion, mixed halide rearrangements—motivated as DFT-intractable at similar length/time scales.
Ensembles and analysis (read from full text). The article motivates NPT/NVT choices for thermal exploration of perovskite phase behavior and octahedral tilt order parameters; exact thermostat/barostat settings, timestep, and equilibration segments should be copied from Methods in the version-of-record PDF rather than inferred from this summary. Post-processing typically tracks lattice metrics, pair correlations, and order-parameter thresholds for cubic-like symmetry versus tetragonal/orthorhombic motifs.
MD application (reactive, large cells). Production runs use LAMMPS with ReaxFF (see the article) for halide perovskite CsPb(Br\(_x\)I\(_{1-x}\))\(_3\) supercells at sizes and stoichiometries in Methods; N/A on this page for exact atom counts. Periodic boundary conditions apply. NVT/NPT as stated for temperature sweeps; N/A—exact NVT/NPT choice, time step (fs), equilibration/production duration (ps/ns), and Nosé–Hoover (or other) thermostat parameters are taken from the J. Phys. Chem. C text, not the galley stub. N/A—no external electric field; N/A—no umbrella/metadynamics; N/A—pressure-tensor targets only if NPT is used—see VOR for hydrostatic pressure and barostat settings.
Findings¶
Increasing Br content lowers the temperature at which the structure attains a cubic-like average symmetry in the simulations. Br for I substitution introduces local strain (smaller Br⁻ radius) that couples to octahedral tilting/rotation dynamics; the perturbation is reported to remain cooperative over ~2 nm, rationalizing why low Br fractions (e.g., x ≤ ¼ in CsPb(BrₓI₁₋ₓ)₃) strongly affect phase stability despite dilute substitution.
Interpretive point: long-ranged correlations in octahedral motion mean local chemistry (Br vs I) can shift global phase behavior—an argument used to connect atomistic trajectories to device-relevant thermal windows (discussion as summarized in the article). - Training breadth: expanding from CsPbI₃ to CsPbBr₃ and mixed halides requires additional DFT targets for halide sub lattice energetics and competing non-perovskite phases so the reactive model remains transferable across x (Methods/SI overview).
Limitations¶
Full simulation protocols (ensemble choices, thermostat/barostat, run lengths, and quantitative transition temperatures) should be read from the Methods section of the full PDF.
Operational note: when comparing to experiment, map simulated order parameters for cubic-like symmetry to diffraction or spectroscopic fingerprints in the article, rather than assuming bulk Tc from finite cells alone.
Relevance to group¶
Core ReaxFF extension for halide perovskites with explicit DFT training sets and phase-behavior application, aligned with the group’s parameterization pipeline.
Citations and evidence anchors¶
- DOI:
10.1021/acs.jpcc.4c00563