Quantitative atomic force microscopy: A statistical treatment of high-speed AFM data for quality control applications

Prior to the recent developments of high-speed atomic force microscopy (HS-AFM), atomic force microscopy (AFM) was not favoured by industry because of its complexity and slow image acquisition speed which may lead to poor resolution and unreliable quantified results. HS-AFM, however, is capable of imaging several frames per second and thus capable of quick, accurate, and representative measurements - ideal for quality control applications. This study demonstrates HS-AFM as a useful quality control tool by assessing the roughness of silicon carbide (SiC) monofilament fibres as a challenging example of how large HS-AFM datasets in excess of 200 images can be collected and used for reliable quantification. A comparison of two roughness methods utilising either area or line roughness parameters has been conducted, where very little difference was noted apart from the lower statistical significance of line roughness. The roughness of ten SiC fibre samples was measured with Sa (the area equivalent to Ra) roughness results varying from 34 nm to 53 nm. The small measurement uncertainties, as a result of the large number of measurements, meant that the roughness results were distinguishable from one another even though all ten SiC fibres were very similar to each other, having been produced by the same manufacturer and process. The robustness of the methods have been tested by repeating the analysis for each fibre, and in the one instance where the repeated data did not agree, a further dataset proved which one was incorrect, illustrating how industry can use these methods for quality control. A methodology of identifying the minimum number of frames required to account for sample variability, as well as recommendations on how to use HS-AFM for quantitative measurements in quality control, are also included to enable easy reproduction and adaptation of this work for other samples and measurements.