Fast focus-based depth detection for manipulation in scanning electron microscopes

To date, determining depth information for robotic handling in a scanning electron microscope is a challenging problem without a versatile solution. In this paper, focus-based as well as other depth detection methods are analyzed showing limitations in terms of accuracy and speed. To overcome these limitations, existing focus-based approaches are combined with a new line-scan based position tracking in order to increase both accuarcy and update rate. Using normalized variance as focus measure on each line scan results in well-defined focus curves. The depth from defocus approach exploits the linearity of the normalized variance for defocused objects to facilitate fast coarse positioning. The depth from focus approach determines the peak of the normalized variance during a focus sweep with high accuracy and enables fine positioning using a look-then-move methodology. For the first time, a z-alignment on the nanoscale is implemented solely relying on the SEM's focus information.