| Abstract Scope |
In the spirit of Chris Wolverton’s integration of first-principles theory, thermodynamics, and data-driven design, this talk asks whether electron microscopy can become a self-driving, theory-aware characterization engine: an experimental counterpart to the high-throughput computation it informs. Prediction of stability and properties has been transformed by first-principles databases and machine learning, yet the experimental loop that must confirm, contradict, or extend those predictions remains rate-limiting.
We describe strategies that ration electrons and time across large composition and microstructure spaces by coupling automated acquisition, machine-learning interpretation, multimodal imaging, diffraction, spectroscopy, and data provenance. Examples include Transmission Kikuchi Diffraction for rapid symmetry and space-group assessment, multimodal TKD/EDS mapping of megalibrary nanoparticles, nD-STEM workflows that unite diffraction, imaging, and spectroscopy, and nanofabricated in-situ stages for dynamic processes.
The aim is closed-loop hypothesis testing that extracts provenance-anchored descriptors linking structure, chemistry, stability, computation, synthesis, and phase-stability models. |