Atomic Force Microscopy (AFM) lies centrally within developments in nanotechnology without material restrictions and is increasingly being used for nanoscale characterization in a wide variety of physical, biological and chemical processes. In general for AFM, the tip itself has often been the barrier to translating atomic (and beyond) resolution into physical understanding, with many images and processes ultimately being identified as a convolution with the tip structure. Recently, the use of a relatively inert functionalized tip means it can approach very close to the object of interest, allowing the interaction to be dominated by extremely short-range Pauli repulsion between atoms in the sample and at the tip apex - this provides the very high resolution at the heart of this new approach. However, once the technique moves into the much wider world of three-dimensional molecular structures, the link between image and interpretation becomes much more complex, and cannot be elucidated from current modelling approaches. This is a significant barrier to the wider adoption of AFM in molecular characterisation, and prevents its obvious potential being fully realised.
The CATAFM project targets an opportunity to develop a systematic software approach to understand and predict AFM images for molecules of any size, configuration or orientation. We use the latest modelling approaches to build a database of 3D AFM images for a wide variety of molecular structures - effectively a computational tomographic approach. This is integrated into a machine learning infrastructure that can then predict molecular structure directly from an arbitrary AFM experimental image, without any of the current constraints on dimensionality and shape. This opens the door to apply this powerful technique to a huge variety of systems where routine atomic and chemical structural resolution can be a major breakthrough.