Skip to content

What Happens at Surfaces and Grain Boundaries of Halide Perovskites: Insights from Reactive Molecular Dynamics Simulations of CsPbI3

Evidence and attribution

Authority of statements

Prose sections below (Summary, Methods, Findings, etc.) are curated summaries of the publication identified by doi, title, and pdf_path in the front matter above. They are not new primary claims by this wiki.

For definitive numerical values, reaction schemes, and interpretations, use the peer-reviewed article (and optional records under normalized/papers/ when present)—not this page alone.

Summary

ReaxFF MD study of CsPbI\(_3\) focusing on how surfaces, surface defects, and grain boundaries participate in degradation chemistry and relative stability. The study relate computed surface stability trends to experimental prevalence of facets where comparable data exist, and mechanistically track evolution of PbI\(_x\)-like local coordination environments via octahedral connectivity changes (corner → edge → face sharing). Pb dangling bonds and iodine sterics are highlighted as drivers of degradation reactions; defect engineering can stabilize some boundaries by increasing steric hindrance even though clustered defects often accelerate failure.

Methods

1 — MD application (atomistic dynamics)

  • Engine / code: LAMMPS with the authors’ CsPbI\(_3\) ReaxFF (prior ref 33 in the article) using dynamical bond order for bond making/breaking.
  • System size & composition: Orthorhombic CsPbI\(_3\) slab models for (110), (020), and (202) faces with stoichiometric, Pb-poor, and Pb-rich terminations; (110) slabs with 4–10 octahedral layer thicknesses for phase core–shell analysis; cubic-phase Σ grain boundaries 3Σ(112)(0.4,0), 5Σ(210)(0.4,0), 3Σ(111)(0,0) (see SI for lattice vectors and build recipes).
  • Boundaries / periodicity: 3D PBC slabs; GB cells as bicrystals in the cubic description in SI.
  • Ensemble / barostat / pressure / field / enhanced sampling: NVT-style constant-T thermal ramping and holds for the degradation studies; N/A — no NPT stress-control focus; N/A — no external electric field; N/A — no metadynamics (standard ReaxFF MD).
  • Timestep: N/A in the first main-text pages parsed here; use SI section 1 for integration settings when reproducing (typical ReaxFF 0.25 fs is not asserted in the excerpted text).
  • Temperature / duration: Surface onset scans from 300 K to 700 K in 50 K steps; some (110) surfaces run up to 5 ns at 700 K in the narrative; degradation dynamics at 600 K illustrated in Figures 3–4; GB snapshots at 600 K after 200 ps (Figure 6) and twin 3Σ(111) trajectories for 2 ns without degradation; thermostat details in SI.

2 — Force-field training (context)

  • N/A as a new fit in this article — the paper uses the parametrization from ref 33; validation tables in this SI /main text compare ReaxFF to DFT where cited (e.g. valence angles in §2.2).

3 — Static QM and experiments

  • DFT: Single-point or cluster DFT angles (148°/180°) used to label orthorhombic vs cubic character (Figure 2 caption text); not a PES study of reaction barriers here.
  • Experiments (literature): XRD facet prevalence and TEM degradation at GBs from cited work for qualitative agreement with the simulated stability trend — no new lab data in this MD paper.

Findings

Surfaces: A stability ranking (110) > (020) > (202) for orthorhombic slabs under 300–700 K thermal loading matches the relative XRD occurrence trend the authors quote. Degradation proceeds by rearranging PbI\(_x\) octahedra corner → edge → face sharing; Frenkel-like I defects precede PbxI\(_y\) clustering and RDF changes (e.g. Pb–Pb peaks at ~4.2 Å on Pb-rich (110) at 600 K). Pb dangling bonds and low I steric hindrance are argued to promote failure; Pb-poor (110) can be very stable (up to 700 K in the 5 ns case described for that termination).

Grain boundaries: Some GBs degrade (amorphous PbxI\(_y\) patches after hundreds of ps); the 3Σ(111) twin remains intact over 2 ns at 600 K because Cs in cavities sterically blocks octahedral motion — the mechanistic theme is defect/GB geometry gating I mobility and Pb clustering. Comparisons to laboratory work are indirect — the (110)>(020)>(202) trend is said to agree with which facets are prevalent in XRD (Methods+body+citations+[37]–[40], in the VOR PDF). Temperature is a major sensitivity (300–700 K scans, 600 K GBs+2 ns / 5 ns (110) holds). Limitations (authors+ReaxFF+KB): Classical RMD (no carriers, no humidity+optics in the base model); extract+VOR+2022pols-venue-paper SI for lattice replicas+RDFs+tolerances.

Limitations

Classical reactive model for complex electronic-structure-sensitive photophysical systems; long-time annealing and photogenerated carriers are outside the classical ReaxFF scope unless augmented by higher-level calibration. Device humidity, halide migration, and organic cation loss are multiphysics concerns that typically require complementary continuum or electronic-structure studies beyond the inorganic surface focus summarized here. Grain boundary structures in the simulations are idealized; experimental films contain tilt boundaries and impurity segregation that can dominate long-term degradation.

Relevance to group

Connects the group’s ReaxFF expertise to halide perovskite durability at interfaces and microstructure features that dominate practical thin-film devices.

Citations and evidence anchors

https://doi.org/10.1021/acsami.2c09239 — Abstract (~p. 1) states mechanistic claims about octahedral sharing and steric control.