Revealing intrinsic vortex-core states in Fe-based superconductors through machine-learning-driven discovery

Electronic states within superconducting vortices hold crucial information about paring mechanisms and topology. While scanning tunneling microscopy/spectroscopy(STM/S) can image the vortices, it is difficult to isolate the intrinsic electronic states from extrinsic effects like subsurface defects and disorders. We combine STM/S with unsupervised machine learning to develop a method for screening out the vortices pinned by embedded disorder in Fe-based superconductors. The approach provides an unbiased way to reveal intrinsic vortex-core states and may address puzzles on Majorana zero modes.