A High-speed AFM Scanning Mode Based on Learning Control

Abstract

Atomic force microscopy (AFM) is a main instrument for nano-scale measurement and manipulation. This paper proposes a learning control based high-speed scanning mode for an AFM system. Specifically, a learning-based control scheme is designed for the AFM system, which consists of an optimal inverse compensator for the AFM scanner dynamics, and a learning algorithm attacking the surface profile of the detected sample. Then, based on the observation of the offset among neighboring scanning lines, the aforementioned learning-based control scheme is combined with a conventional proportional-integral (PI) controller to achieve a high-speed AFM scanning mode. For periodic samples, this mode can be utilized to largely increase the measurement speed and precision, and simultaneously maintains the distance between the cantilever tip and the detected sample within a reasonable range to avoid the possible harm to them. Therefore, the proposed high-speed scanning mode can be employed for on-line inspection of fast biologic processes, and it can also be utilized to implement such nano-manipulation as repetitive writing.