Logo of the Physikalisch-Technische Bundesanstalt
Production sequence of Si-spheres and interferometrical determination of the sphere volume

Comparison of areal roughness measurements by atomic force and confocal microscopy


The roughness is decisive for functional properties of surfaces such as the friction of components in machines, the distribution of lubricants, the adhesion and, last not least, how a surface appears to the human eye. Consequently, roughness measurements belong to the most important and most frequently performed dimensional measurements in industrial quality assurance. Already for decades, stylus profilometers are used in industrial practice for this purpose, and roughness determination by profilometry is harmonized by a set of standards for long. Its resolution, however, comes to a limit with the smallest conventional tip radius of 2 µm. Furthermore, many measurement tasks on technical surfaces demand an areal measurement of textures.

Optical surface measurement techniques such as interference and confocal microscopy acquire areal measurements and, additionally, they are fast and non-contacting, which leads to increased applications in industry. For several years, the standards series ISO 25178 is being developed for areal surface texture measurements. At the same time, profilometric roughness measurements by stylus instruments are revised accordingly and put into a new order by the standards series ISO 21920, which will be published soon.

Practical experience shows that optical roughness measurements can be impaired by a number of optical effects, which do not only depend on the measurement method, but also on instrument properties such as the optical components used, the user settings chosen for the measurement and the algorithms applied to process the signals in the instrument. This complexity may overcharge many users, thereby aggravate the target-oriented application of these methods and thus impede the comparability of results (see also report on EURAMET comparison 1242). Without additional information, the user often cannot safely identify all disturbing influences.

Systematic deviations in optical surface measurements can be revealed by comparison with atomic force microscopy (AFM), which provides a higher spatial resolution. This is most successful when measuring at the same location with both instruments. For this purpose, PTB was involved in the development of transfer standards with marked reference fields.

Fig. 1 shows, as an example, a standard of the type UFRS (Ultrafine Roughness Standard). Such standards are produced by focused ion beams (FIB; manufacturer: company point electronic, Halle/Saale, Germany). The advantage of this technology is that a given model topography can be transferred point by point. This does not only allow to write much smaller structures with significantly steeper slopes, but generally allows for much more flexibility than the conventional mechanical or chemical techniques classically used for surface structuring. As input, a mathematically computed or a measured – and possibly modified – dataset can be used. The UFRS profiles are based on downscaled synthetic profiles otherwise used for superfine roughness standards (SFRS). Here the structured field is sized 145 µm x 145 µm, surrounded by a frame for better orientation. This ensures that different instruments manage to measure at the same location. The design value for Sz is 75 nm and for Sa 10 nm. Fig. 1 shows the comparison of measurements by a confocal laser-scanning microscope (CLSM) and an AFM. A closer look at both images reveals that the AFM image appears a bit sharper, and in the CLSM image a longwave background can be seen that indicates a flatness error. Both histograms (frequency of occurance of the measured height values) generally agree fairly well, but the histogram of the AFM measurement depicts some discrete peaks. These peaks originate from the FIB process that is applied in layer-by-layer mode in 16 discrete steps. The discrete height levels get completely blurred in the CLSM measurement.

Such standards do not only allow to compare the roughness values obtained with different methods, as shown here in Fig. 1, but also to determine e. g. the measured distribution of slope angles and the spatial frequencies (FFT, PSD) contained in the measured images.

Fig. 1: The same field of 65 µm x 65 µm on an UFRS, measured by Olympus LEXT CLSM (left, 100x objective used, AN = 0.95) and SIS AFM (centre) with same colour-coding of height (black-to-white equals 50 nm), 1024 x 1024 pixels each, and the corresponding height histograms (right)

Tab 1: Comparison of CLSM and AFM roughness values for the images shown in Fig. 1. Ssk and Sku dimensionless, all others in nm. Images levelled, not filtered. Roughness analysis by software SPIP (Image Metrology A/S, Denmark). The parameter Sδ5-95 is the height difference between the 5 % and the 95 % intersection levels in the material ratio curve according to ISO 4287, i. e. the height interval containing the central 90 % of the measured height values in the histogram of height values, while both the highest and the lowest 5 % of height values, which are often caused or influenced by measurement artefacts, are excluded from this amplitude parameter.

LEXT CLSM9.612.00.242.973.114.730.78.740.6
SIS AFM8.710.70.152.761.610.929.16.935.1





Physikalisch-Technische Bundesanstalt
Bundesallee 100
38116 Braunschweig