Researchers achieve 13x precision boost in electron microscopy

Researchers from Eindhoven University of Technology (J.S. van Hulst et al.) have developed an advanced control framework that dramatically improves specimen positioning precision in electron microscopy. Published on arXiv (2608.03669), their work targets the persistent challenge of maintaining nanometer-scale positioning accuracy over long strokes in operational electron microscopes.

The team's integrated solution combines three key innovations: per-element hysteresis compensation to linearize actuator responses, vision-based sensing that constructs POI measurements from electron microscope images via cross-correlation, and commutation-angle-domain iterative learning control that cancels repeatable disturbances. This framework achieved error reductions of 13x in lab settings and 7-12x in actual transmission electron microscope operations, where dedicated POI sensors are typically unavailable. The approach parameterizes learned corrections in the commutation angle domain, enabling transfer across different drive frequencies while maintaining quasi-static positioning accuracy.