Characterisation of structured surfaces and assessment of associated measurement uncertainty

Abstract

Recently, structured surfaces, consisting of deterministic features designed to produce a particular effect, have shown promise in providing superior functional performance for a range of applications including: low friction surfaces, hydrophobic surfaces and optical effects. Methods have been developed to characterise such structured surfaces. The most widely used characterisation methods are based on segmenting the surface in feature and background regions and then determining the geometrical properties of those features. However, further work is needed to refine these characterisation techniques and provide associated uncertainties. This thesis considers the effect of various segmentation control parameters such as thresholds on the final geometric parameters. The effect of varying filter size is also considered. These considerations should help in selecting a suitable characterisation method for future projects. Additionally, uncertainty in the characterisation should be estimated in order to give an indication of the accuracy of the assessment. However, no previous work has assessed uncertainty in the dimensional properties of structured surfaces. Therefore, this thesis presents two methods to characterise the uncertainty in the geometric characteristics of structured surfaces. First, the measurement reproducibility is used, which can be determined by repeated measurement of a feature. However, measurement reproducibility cannot account for all sources of uncertainty and cannot assess any bias in the measurements. Therefore, a second method based on assessment of the metrological characteristics of the instrument is considered. The metrological characteristics estimate errors produced by the instrument in a way that can easily be measured. Monte Carlo techniques are then used to propagate the effects of the metrological characteristics and their uncertainties into the final measurement uncertainty. For the example used, it was found that the results using the metrological characteristics were in good agreement with the reproducibility results. From these results, it is concluded that the choice of segmentation method, control parameters and filtering can all significantly effect the characterisation of features on a structured surface, often in unexpected ways. Therefore, care must be taken when selecting these values for a specific application. Additionally, two methods of determining the uncertainty of the structured surfaces were considered. Both methods are valid and produce similar results. Using the measurement reproducibility is simple to perform, but requires many measurements and cannot account for some uncertainty sources such as those due to the instrument amplification factors. On the other hand, the use of metrological characteristics can account for all significant sources of uncertainty in a measurement, but is mathematically more complex, requiring Monte Carlo simulations to propagate the uncertainties into the final characteristics. Additionally, other artefacts than the sample being measured are required to determine the metrological characteristics, which may be an issue in some cases.