Digital signal processing in AFM topography and recognition imaging

Atomic force microscopy (AFM) has proven to be a powerful tool to observe topographical details at the nano- and subnanometer scale. Since this is a rather new technique, new enhancements with faster scanning rates, more accurate measurements and more detailed information were developed. This requires also a higher demand on the signal processing and the controlling software. Operating an AFM with analog driven hardware is often limited by drift and noise problems. Here we overcome this problem by introducing digital signal processing capable of accurately stabilizing the piezo control in the newly developed TREC (topography and recognition imaging) mode. In this mode topographical information and molecular recognition between tip bound ligand and surface bound receptors is simultaneously acquired. The sought information is conveyed by slight variations of the minima and maxima of the signal amplitudes. These variations are very small compared to the maximum possible DC deflection. Furthermore, the DC offset exhibits a rather large drift mostly attributed to temperature changes. To obtain reliable tracking results the oscillating photodiode signal needs to be nonlinearly filtered and efficiently separated into four major components: the maxima, the minima, the spatial average of the maxima, and the spatial average of the minima. The recognition image is then obtained by a nonlinear combination of these four components evaluated at spatial locations derived from the zero-crossings of the differentiated signal resulting from a modified differentiator FIR filter. Furthermore, to reliably estimate the DC drift an exponential tracking of the extrema by a first-order IIR filter is performed. The applicability of the proposed algorithms is demonstrated for biotin and avidin.