AI–Assisted Rapid Identification of Impurity Phases and Magnetic Ordering in Neutron Diffraction

We introduce an automated workflow that resolves unexpected peaks in neutron diffraction by proposing impurity phases, testing symmetry-allowed magnetic subgroups, and returning a standardized refinement project. Given a user-specified primary phase, the system mines a curated catalog for impurity hypotheses, performs first-pass refinements, and programmatically queries the Bilbao Crystallographic Server for subgroup generation. Embedded machine-learning models guide hypothesis ranking and search heuristics by leveraging features from peak positions, intensities, and refinement residuals, reducing manual trial-and-error. The approach supports time-of-flight and constant-wavelength data and is designed for rapid operator-in-the-loop feedback. Case studies demonstrate faster triage of "mystery peaks" and smoother hand-off to expert refinement, providing a practical step toward autonomous, AI-assisted neutron experiments.