Jörg Seewig

Areal Filtering Methods

Filtering is essential for surface texture characterisation. Filtration separates the measured data into different scales of interest. The extracted scales can be characterised by parameters given in ISO 25178 part 2. In this chapter the two draft areal filtering standards, ISO 16610 part 61 (areal Gaussian filter) and ISO 16610 part 71 (robust areal Gaussian regression filter), are discussed. Spline, morphological and wavelet filters are also briefly introduced.

Extraction of shape and roughness using scattering light

The extraction of 3D shape and roughness by optical measurement techniques become more and more import in industrial applications. Optical systems are measuring fast with high accuracy and give reliable information about the workpiece form or surface roughness. The classical systems based on triangulation, white light, confocal, shadow or fringe projection techniques and are applied with a great success in recent years. In future there will be a growing interest in robust inline measurement techniques to monitor the manufacturing process. E. g. some automotive manufactures are using confocal systems to characterize the surface of cylinder liners inline. But there is another robust and powerful technique suitable for inline measurement purposes: scattered light sensors. In this paper, a special type of a scattered light sensor based on the former Rodenstock RM 400 sensor is considered. The sensor enables the user to measure form and roughness in a robust manner. The properties of the sensor are analyzed system-theoretically.

Novel parameters to assess feature size on functional surfaces

In surface metrology it is often beneficial to characterize the size of features on a topography in a descriptive way. Granulometry is one well known method in image processing that offers information about the size distribution of binary and greyscale images. Here the granulometry of micro-structured functional surfaces is explained and applied to topographical measurements from a confocal scanning microscope. The surface was laser textured and exhibits regular circular features. A set of parameters is derived from the size distribution. The proposed parameters correspond to the lateral and vertical dimension of the features.

Calibration of z-axis linearity for arbitrary optical topography measuring instruments

The calibration of the height axis of optical topography measurement instruments is essential for reliable topography measurements. A state of the art technology for the calibration of the linearity and amplification of the z-axis is the use of step height artefacts. However, a proper calibration requires numerous step heights at different positions within the measurement range. The procedure is extensive and uses artificial surface structures that are not related to real measurement tasks. Concerning these limitations, approaches should to be developed that work for arbitrary topography measurement devices and require little effort. Hence, we propose calibration artefacts which are based on the 3D-Abbott-Curve and image desired surface characteristics. Further, real geometric structures are used as an initial point of the calibration artefact. Based on these considerations, an algorithm is introduced which transforms an arbitrary measured surface into a measurement artefact for the z-axis linearity. The method works both for profiles and topographies. For considering effects of manufacturing, measuring, and evaluation an iterative approach is chosen. The mathematical impact of these processes can be calculated with morphological signal processing. The artefact is manufactured with 3D laser lithography and characterized with different optical measurement devices. An introduced calibration routine can calibrate the entire z-axis-range within one measurement and minimizes the required effort. With the results it is possible to locate potential linearity deviations and to adjust the z-axis. Results of different optical measurement principles are compared in order to evaluate the capabilities of the new artefact.

Aktive Schwingungsdämpfung eines Weißlichtinterferometers

Zusammenfassung In der Oberflächenmesstechnik sind mechanische Umgebungsschwingungen problematisch. Eine aktive Dämpfung von Messgeräten kann aus diesem Grund ein wichtiger Bestandteil für eine hohe Ergebnisgüte sein. Beispielhaft ist hier die praktische Umsetzung der Dämpfung eines Weißlichtinterferometers (WLI) vorgestellt. Als Referenzmessgerät dient ein Quadraturinterferometer (QI), welches die Schwingungen zwischen WLI-Messkopf und Messtisch aufzeichnet. Ein Regelalgorithmus nutzt diese Information um eine Stellgröße zu erzeugen, welche den WLI-Messkopf mit Hilfe eines Piezoelementes stabilisiert. Abstract Environmental vibrations are always problematic in surface metrology. For this reason an active damping of measurement setups is an important component for a high measurement quality. As an example the practical realization of the damping of a White Light Interferometer (WLI) is presented here. A quadrature interferometer (QI) serves as reference measurement device which detects the vibrations between the measurement head and the measurement table. A closed loop control algorithm uses this information and delivers an actuating signal for the stabilization of the measurement head by means of a piezoelectric element.

Robust evaluation of intensity curves measured by confocal microscopies

Confocal microscopy is a state of the art optical principle to measure the topography of technical surfaces. The output of the measuring process is an intensity curve for each scanned point on the topography. The maximum of the intensity curve correlates to the point height. However, the intensity function is influenced by the geometrical properties of the surface topography and its material. Simple peak picking of the intensity curve leads to insufficient results when calculating the point height. Therefore, the centre-of-gravity or fitting algorithms are preferred. Both have an integral behaviour and are able to suppress unwanted signal components. Today, the centre-of-gravity is often the state of the art method. Disadvantages are e.g.: the method is sensitive against vibrations of the instrument during the measuring process. In contrast to the centre-of-gravity calculation, fitting algorithms are numerically inefficient and slow. Moreover, the fitting process needs a priori information about the curve of the intensity function. We propose an alternative algorithm for the robust evaluation of intensity curves. The amplitude spectra of each intensity curve of a measured reference data set are analysed. The applied technique is based on the Cramér-Rao bound and leads to a threshold operator for calculating the centre-of-gravity in the frequency domain. A possible phase distortion (an asymmetrically shape of the intensity function) caused by diffraction or optical aberrations will also be significantly suppressed. The performance of the algorithm is shown and we compare the algorithm to the popular centre-of-gravity.

Application of ordinary kriging for interpolation of micro-structured technical surfaces

Kriging is an interpolation technique used in geostatistics. In this paper we present kriging applied in the field of three-dimensional optical surface metrology. Technical surfaces are not always optically cooperative, meaning that measurements of technical surfaces contain invalid data points because of different effects. These data points need to be interpolated to obtain a complete area in order to fulfil further processing. We present an elementary type of kriging, known as ordinary kriging, and apply it to interpolate measurements of different technical surfaces containing different kinds of realistic defects. The result of the interpolation with kriging is compared to six common interpolation techniques: nearest neighbour, natural neighbour, inverse distance to a power, triangulation with linear interpolation, modified Shepards method and radial basis function. In order to quantify the results of different interpolations, the topographies are compared to defect-free reference topographies. Kriging is derived from a stochastic model that suggests providing an unbiased, linear estimation with a minimized error variance. The estimation with kriging is based on a preceding statistical analysis of the spatial structure of the surface. This comprises the choice and adaptation of specific models of spatial continuity. In contrast to common methods, kriging furthermore considers specific anisotropy in the data and adopts the interpolation accordingly. The gained benefit requires some additional effort in preparation and makes the overall estimation more time-consuming than common methods. However, the adaptation to the data makes this method very flexible and accurate.

Light Scattering Methods

Light scattering belongs to a class of techniques known as area-integrating methods for measuring surface texture. Rather than relying on coordinate measurements of surface points, light scattering methods probe an area of the surface and yield parameters that are characteristic of the texture of the area as a whole. The specular beam intensity, the angle-resolved scatter and the angle-integrated scatter are examples of measurands from light scattering that can yield useful parameters of the surface texture. Uses of light scatter for inspecting the surfaces of mechanical and optical components as well as surfaces produced in semiconductor manufacturing are primarily reviewed here. Several documentary standards describing best practice are also briefly reviewed.

Uncertainty of height information in coherence scanning interferometry

Coherence scanning interferometry CSI with a broadband light source (short known as white light interferometry) is, beside the confocal technique, one of the most popular optical principles to measure surface topography. Compared to coherent interferometry, the broadband light source leads, theoretically, to an unambiguous phase information. The paper describes the properties of the correlogram in the spatial and in the frequency domain. All deviations from the ideal correlogram are expressed by an addition phase term. The uncertainty of height information is discussed for both, the frequency domain analyse (FDA) proposed by de Groot and the Hilbert transform. For the frequency domain analyse, the uncertainty is quantified by the Cramér-Rao bound. The second part of the paper deals with the phase evaluation of the correlogram, which is necessary to achieve a high vertical resolution. Because the envelope function is often distorted, phase jumps lead to ambiguous height informations. In particular, this effect can be observed measuring rough surfaces.

Kenngrößen der Abbott-Kurve zur integralen Beurteilung dreidimensional gemessener Zylinderlaufbahn-Oberflächen

In diesem Bericht der Universitat Hannover werden Kenngrosen vorgestellt, die die Hohe und Form der Abbott-Kurve dreidimensional gemessener Oberflachenausschnitte kennzeichnen. Diese Kenngrosen erlauben eine integrale, funktionsbezogene Beurteilung von Zylinderlaufflachen hinsichtlich ihres Einlaufverhaltens, mechanischer Belastbarkeit und ihrer Olaufnahme. Die neu entwickelte Auswertmethodik wurde in Zusammenarbeit mit der deutschen Automobilindustrie im Arbeitskreis 3D-Rauheitsmesstechnik entwickelt und erprobt. Zur internationalen Normung sind diese Kenngrosen bereits in die ISO eingebracht worden.