Abstract

Atomic force microscopy (AFM) is an imaging technique which can measure the surface structure of a specimen of interest down to nano-scale. It does this by scanning a tiny probe across the surface somewhat like a record player pick-up and thereby measuring a “height map” or other properties of the surface.

In an ongoing research project at Aalborg University we are attempting to speed up imaging or reduce interaction with the sample in AFM . We seek to achieve this by scanning the surface sparsely and thereby reducing the scan time and/or reducing the fraction of the surface area being affected by the probe. This naturally poses some challenges in the imaging process due to the resulting “missing data”.

In this talk we explain how we use computational techniques inspired by compressed sensing, optimisation and inverse problems to reconstruct AFM images from the otherwise insufficient measurements. We explain how we search for and optimise sampling patterns to be better suited for subsequent reconstruction of the images. We present some challenges met in the process of developing the proposed techniques from synthetic numerical experiments to real data from commercial AFM equipment and show examples of the results obtained so far.