Contour-oriented automatic tracking based on Gaussian processes for atomic force microscopy

Abstract A local scan method based on Gaussian processes inference was developed for efficient atomic force microscopy measurements of string-like nano-objects such as DNA molecules, nanowires, shape boundaries, and so on. Unlike conventional line-by-line raster scan, the proposed method can adaptively track the nano-objects deposited on a flat substrate through accurate prediction of the x- and y- coordinates along the objective contour. Therefore, the used time in scanning the featureless substrate is drastically eliminated. Compared with conventional raster scan, the time efficiency can be improved by almost one order of magnitude. Numerical simulations and practical experiments demonstrated that such a method has several advantages including automatic tracking, significant reduction of scan distance and improvement of measurement efficiency.