A key enabler of Nanoscience and Nanotechnology is the Atomic Force Microscope (AFM) that has opened up new realms, of interrogation and manipulation of matter at the atomic scale. It has resulted in breakthroughs in understanding sub-atomic molecular structure, protein folding dynamics and materials characterization. The potential impact of significantly faster AFM based imaging is immense, e.g., it will allow the study of dynamics of material at the nanoscale that was hitherto not accessible. The aim of this research is to study modern signal processing techniques that achieve gains in imaging speeds by an order of magnitude. The investigators will push the synergistic transfer of knowhow between the engineering and the AFM communities through student visits and specialized workshops. The findings will be integrated at appropriate levels into the undergraduate and graduate curriculum.
The main component of an AFM is a cantilever that deflects in response to forces at the pico-Newton scale. The focus of this research is the dynamic mode AFM operation, where the cantilever gently taps the sample being imaged; the mode of choice for imaging biological samples. Even though forces do form a good indicator of sample topography, the system memory and the nonlinearities of the AFM system dynamics preclude a direct mapping of the cantilever forces into a finer description of topography, especially at high imaging speeds. The investigators model the complex nanoscale interactions in a mathematically tractable manner that facilitates the development of maximum a posteriori (MAP) sequence detection of the sample features. Factor graph representations and the development of appropriate inference schemes will be studied. The algorithms will be implemented on FPGAs and tested exhaustively with experimental data.
