Estimation of tip-sample interaction in tapping mode AFM using a neural-network approach

In Atomic Force Microscopy, the tip-sample interaction force contains information about the sample topology, material properties and tip geometry. However, quantitative measurement of the time-varying tip-sample interaction forcing function is challenging in the tapping mode because of the combined dynamic complexities of the cantilever and nonlinear complexity of the tip-sample force. In this paper, an initial investigation of a neural-network approach to tip-sample interaction force estimation is studied. The tip-sample interaction is treated as an unknown position-dependent force on the cantilever. A modified radial basis function neural-network is used in a dynamic observer framework to approximate the unknown forcing function. Design of the observer gains is discussed and simulations are used to demonstrate plausibility of the approach. Accuracy of the force model is evaluated for several different tip-sample distances and materials and future direction are discussed.