• AFM
• AFM Image fusion
• Automation
• Image alignment
• Nanometrology
• Nanotechnology
• Roughness
• RTS SPM
• Statistics
• Surface science

Since their first introduction a few decades ago, Scanning Probe Microscopy (SPM) techniques have emerged as extremely powerful tools to investigate material properties at the nanoscale. Atomic Force Microscopy (AFM) is likely the most diffused SPM, thanks to its ability to precisely and reliably map surface topography and other local properties, and the possibility to be used with a virtually unlimited class of materials without the need of prior surface preparation. Nonetheless, its adoption within industrial settings, for instance those dealing with micro- and nano-electronics, is still limited, mostly because its operation requires dedicated and time wasting operator efforts. As a matter of fact, the geometric features of the probe tip heavily affect the quality of the produced images, resulting in the frequent need for tip replacements.
This thesis examines the inherent limitations of AFM, focusing on the critical dependency of data quality on probe tip characteristics. In AFM, tip shape and sharpness are essential for precise surface characterization. Even slight variations or degradation in tip quality can introduce inconsistencies and biases in both topographic measurements and more advanced material property evaluations. However, as tip quality and, consequently, data quality deteriorate due to wear, the need to use multiple tips becomes inevitable. A significant challenge here is the manual tip replacement process, which may take from ten minutes to several hours, thereby limiting the frequency of exchanges and impeding high-throughput measurements and comprehensive statistical analysis of surface properties.
To address this, the viability of the Reverse Tip Sample (RTS) approach is explored. The RTS configuration reverses the positions of the sample and the tip with respect to the standard SPM configuration. Specifically, for AFM, the sample is now mounted on a tipless cantilever, while the tip is integrated into a nanofabricated RTS probe chip which is placed on the microscope's stage. This reversal enables rapid and seamless tip switching since RTS probe chips host not one, but thousands upon thousands of tips which are easily accessible by the sample through navigation with the piezo and / or coarse motors of the microscope. As such, the RTS configuration not only minimizes experimental downtime but also exploits the inherent variability among tips. Each tip, with its unique geometric characteristics, contributes diverse and complementary data. For example, sharp tips capture high-resolution topography, while blunter tips enhance sensitivity to adhesion forces or stabilize electrical contacts. By continuously cycling through numerous tips, RTS provides a richer statistical representation of tip-sample interactions. Altogether, RTS represents a promising technology, but it is still in its early stages.
This work brings further advances to RTS SPM with the integration of image recognition algorithms, which ensure precise tip alignment with sub-nanometer accuracy after each tip exchange. This automated alignment is essential for repeated measurements of the same region of interest (ROI), where even minor positional inaccuracies can significantly affect measurement consistency. The process begins with a reference scan of the ROI, which is then compared to each subsequent trial scan from a new tip. The image alignment algorithm calculates translational offsets that are applied via piezo motors, thereby ensuring nanometer-accurate centering of the ROI. Moreover, two graphical user interface (GUI) software platforms were developed for a commercial atomic force microscope (AFM) equipped with Python-interfacing APIs. The first platform allows flexible, automated execution of AFM scans using different tips from the RTS probe chip, with each scan configured according to arbitrary parameters for specific target regions. The second platform simplifies the process of running identical scans on the same ROI using different tips, thereby enhancing statistical reliability of the data. Both platforms depend on the above-mentioned precise, automated alignment provided by the image recognition system and both substantially reduce setup / measurement time from several of hours to just a few minutes.
Recognizing that tip quality remains a crucial factor even in the RTS configuration, the thesis explores the development of a tip ranking system. An initial exploratory study identifies surface roughness as a promising, easily obtainable parameter for establishing a quality-based hierarchy among the many tips on a probe chip. The use of surface roughness for tip ranking is further investigated using the previously developed software platforms for experimental automation. The investigation uncovered, and was stalled by, several fundamental scientific questions of current relevance in surface science and nanometrology, namely the dependence of surface roughness on tip size, scan size, and other factors. These issues are addressed and investigated through extensive experimental studies. The experimental results are in line with existing results in AFM literature, and serve to confirm some of the weaker ones with a much higher statistical strength. An analytic computer simulation based on offset curves was developed for simulating the surface roughness measurements of arbitrary line profiles by tips of arbitrary size and shape. The simulation independently reproduces several experimental results, and provides further insights on RMS roughness and on its behaviour against various parameters.
The combination of rapid automated tip exchanges, precise algorithm-driven alignment, and the implementation of a tip ranking system collectively advances the capabilities of AFM for high-resolution nanoscale imaging, and also served for the fundamental investigation of scientific phenomena in surface science.