Research On Control Of AFM Based Nano-Imaging And Nano-Manipulation

Since the invention of the Atomic Force Microscope (AFM), it has been widely utilized to explore the research in the fields of biology, material, chemistry as well as physics. Consequently, as one of the most powerful instruments for nano-imaging and nano-manupulation, AFM and its related technologies are regarded as the "eyes" and "hands" of nano-technology. Hence, the AFM research developments are of vital importance to micro/nano systems and have gained lots of attention across the world. True advancements in AFM instrumentation may lead to a breakthrough for the entire micro/nano research area.Although AFM systems have been widely utilized in practice, the related technologies are still far from satisfaction. Specifically, there still exist some challenging problems in current AFM systems: (1) As an imaging tool, AFM based experiments are usually difficult to repeat due to its complex nonlinearities; besides, AFM is more difficult to use than other microscopes; furthermore, the scanning speed of AFM is usually very slow due to its narrow bandwidth. All those drawbacks hinder its further applications for nano-imaging. (2) When an AFM is used for nano-manipulation, it requires some modifications on its hardware and software structure. Particularly, the widely employed piezo-tube scanner is not able to position the tip precisely due to its hysteresis, cross-coupling and other uncertainties. This is one of the main reasons why AFM based nano-manipulation is still of low efficacy and success ratio.Funded by National Natural Science Fundations, this dissertation conducts a thorough survey of AFM related research topics and recent progresses. Based on that, the dissertation focuses on some of those challenging yet key problems, aiming to improve the imaging and manupulation performance of AFM systems.(1) Modeling and Simulation of AFM Systems.Generally, it is extremely complicated to model all the dynamics of AFM systems based on first principles. For a better understanding of AFM systems, in this dissertation all components of AFM systems are modeled or calibrated experimentally, including the piezo-scanner, photo-diode, and high voltage amplifier. For the atomic interactions between the tip and the sample, and the cantilever dynamics, a theoretical modeling and analysis is given. Based on those models, a virtual tapping-mode AFM is constructed in Matlab/Simulink environment. To prove its validity, the classical approach-retract force curve and bi-stable phenomena are explored.(2) Relay Based Auto Tuning of AFM PI Parameters.It is well known that the tuning of AFM PI parameters is a tedious and complicated procedure, especially when some components or the scanning speed are changed. In this dissertation, a relay controller is employed to excite AFM systems, resulting in a stable limit cycle. The amplitude and frequency of the limit cycle are then used to identify the system critical point. Based on Ziegler-Nichols formula, initial PI parameters will be obtained. When the scanning speed is changed, the PI parameters will be adjusted accordingly to avoid possible system oscillations. Experimental results show that the proposed approach can indentify the system variation due to changes of the components. Moreover, better sample topography image can be achieved after auto-tuning the control gains with a fast scanning speed.(3) H_∞Controller Design and Simulation for Tapping Mode AFM.To enhance the noise rejection, the bandwidth and hence the scanning speed of the system, an H_∞controller is designed for tapping mode AFM. Specifically, based on mixed sensitivity method, proper weighting functions are chosen to scale the system error, control input and system output such that the controller is able to damp the resonance frequency of AFM system and eliminate the possible vibration during high speed operation. Several simulated samples are scanned using the H_∞controller. Simulation results show the controller is more capable of tracking those samples and has smaller interaction forces compared to PI algorithm. Besides, the controller is of superior performance considering the noise rejection and robustness against model uncertainty.(4) Nano-Positioning of Tip by Utilizing Piezo-Tube Scanner.Tip nano-positioning is one of the key factors to the success of AFM based nano-manipulation. To alleviate the positioning error caused by piezo-hysteresis, cross-couping and other uncertainties, the following method is proposed: (a) To compensate for the hysteresis of the piezo-tube, the voltage-displacement relation is obtained from image of calibrated gratings. The inverse voltage is computed to compensate the positioning error. (b) For cross-coupling and structural errors, etc, the tip is used to indent the sample, and the true tip positions are estimated from the indentation points matrix. The positioning error will be compensated according to the differences between the desired tip positions and true tip positions. After that, the tip can be positioned roughly around the desired position. (c) The local scanning method is employed to refine the positioning error, which may also come from thermal drift or other uncertainties. Experimental results show that precise nano-positioning can be achieved using this method.(5) Construction of AFM Based Nano-Manipulation Platform.Based on the tip nano-positioning techniques, an AFM system is modified into a nano-manipulation platform. RTLinux is employed to ensure real-time control of the platform. Besides, the platform is easy to integrate new nano-manipulation applications and enables access to real-time cantilever deflection signals. The dissertation conducts various experiments including nano-imprint, indentation and manipulation to prove its validity.