Light-induced phase separation (LIPS) in mixed lead-halide perovskites is one of the major obstacles to high-efficiency perovskite-based tandem solar cells, yet the mechanisms that govern it remain unclear. Using in situ X-ray diffraction on MAPbBr1.8I1.2, we track coexisting phases through their lattice parameters and observe the same kinetics across all samples: phase separation sets in only after a delay, with an iodide-rich phase appearing abruptly at near-pure composition while the residual mixed phase drifts gradually towards bromide. To capture this behaviour, we extend the Kawasaki Hamiltonian, adding separate polaron-induced stress terms for electrons and holes, and evolve the halide lattice by Monte Carlo simulation. The model reproduces both the delay and the asymmetry, neither of which follows from standard Cahn-Hilliard kinetics. Within the model, the stronger coupling of electron polarons in bromide-rich regions dominates the driving force, pointing to electron-phonon coupling as a lever for suppressing LIPS.
Establishing a unified yet system-specific physicochemical framework for glass formation across covalent, ionic, and metallic systems remains a fundamental challenge, as existing models are largely confined to a single glass family or heavily biased toward compositions that have been successfully vitrified. Here we bridge this gap using a success-bias-free dataset of 78,889 compositional entries drawn from glass-forming region maps. Encoding each composition as an elemental-property matrix derived from the Magpie database, we train a scalable XGBoost framework that attains 89.6% classification accuracy on the all-glass dataset, and generalizes robustly to extrapolative elemental substitutions and higher-dimensional spaces. Shapley analysis of all-glass and system-specific models shows that the four glass families share the same set of frustration mechanisms whose relative weights redistribute systematically with bonding character. Across all families, governing descriptors exhibit a bounded, non-monotonic dependence that we term frustration inversion, whereby compositional heterogeneity stabilizes the amorphous state only up to a family-specific threshold that narrows as network rigidity decreases, furnishing quantitative bounds on the confusion principle. The analysis further identifies BCC-polymorph-referenced elemental descriptors as compositional-level predictors that encode structural preferences toward vitrification. These results establish a cross-system foundation for amorphous stability and a data-driven route to functional glass design.
Existing FAIR materials repositories can only serve pre-computed entries, so broad coverage is perpetually incomplete and cannot adapt to new questions on demand. We present OptiMat Alloys, a large language model-powered conversational agent for multi-principal element alloy exploration. Its central contribution is LLM-based workflow orchestration coupled to a living database: every calculation is performed on demand and persisted automatically with provenance in a local database, so asking a question produces provenance-rich, FAIR-aligned records without a separate deposition step. The agent is accessible through a web interface that requires no programming expertise (low-barrier accessibility) and reports uncertainty through cross-potential and cross-configuration validation (built-in uncertainty quantification). By coupling foundational machine learning interatomic potentials that cover nearly the entire periodic table with natural-language interaction, OptiMat Alloys enables targeted, on-demand computation guided by the user’s domain knowledge and brings computational alloy screening within reach of any materials scientist.
Understanding the microscopic origin of anomalous self-diffusion near melting remains a central challenge in bcc metals. Here we combine ab initio and machine-learning interatomic-potential molecular dynamics (AIMD, MLIP-MD) to elucidate high-temperature mass transport in group-3–6 metals. Comparative AIMD simulations near melting at ambient pressure reveal a marked difference in mass-transport behavior between dynamically-stabilized group-3 and group-4 bcc metals and group-5 and group-6 metals; for which bcc is the intrinsically stable structure. In the former, characterized by strongly anharmonic energy landscapes, sufficiently large thermal displacements of individual atoms trigger cascades of cooperative transport, including ring exchanges and spontaneous formation and migration of vacancy–crowdion-like Frenkel pairs. In contrast, diffusion in group-5 and group-6 metals remains dominated by conventional vacancy mechanisms. Temperature-dependent MLIP-MD simulations of bcc Ti further show Arrhenius-like migration of isolated point defects upon cooling toward dynamical instability. Near melting, collective reactions in nominally pristine lattice regions provide an additional transport channel contributing to non-Arrhenius self-diffusion. Structural analysis reveals transient ω-like local environments as fingerprints of lattice softening accompanying these correlated dynamics. These results identify strong lattice anharmonicity as a key factor promoting collective transport in bcc metals and suggest a broader framework for high-temperature transport in crystalline solids.
We present an AI-driven framework that helps describe and predict damage at the atomistic level in irradiated materials. Rather than identifying discrete, visible defects, we define damage as a continuous spectrum of lattice distortions arising from atomic fluctuations. These distortions are encoded into a compact distortion state vector that unifies information from atomistic simulations and synthetic scanning transmission electron microscopy (STEM). Deep-learning models learn to reconstruct this state directly from STEM images, linking experimental contrast to the underlying atomic disorder without annotations or expert intervention. Operating in this representation space, the framework can forecast the evolution of hidden defect populations from only a few dozen initial images. Thus, it opens new perspectives for understanding how atomic-scale fluctuations give rise to macroscopic degradation, laying the foundation for physics-based digital twins of materials.
