| Abstract Scope |
Despite the ubiquitous presence of microscopy in numerous scientific fields, traditional operations have been largely limited by a manual, human-centric approach. To address these limitations, we developed a cross-platform application program interfaces (API) AEcroscopy (short for Automated Experiments in Microscopy), which is compatible with a variety of vendor devices like atomic force, scanning tunneling, and electron microscopes for automated experimentation via Python-based workflows; it enhances experiment efficiency and reproducibility. The synergy of large language models with AEcroscopy allows seamless conversion of expert concepts into executable codes for microscopy experiments and basic data analyses from AEcroscopy outputs. We further integrate machine learning, such as active learning approach, into automated workflows in scanning probe microscopy to explore ferroelectric and photovoltaic materials, unraveling nanoscale structure-property relationships and physical mechanisms. While our methodologies were initially applied to specific materials, they can be applied in broad synthesis and characterization experiments. |