Measurement noise as a component of uncertainty measurements of the surface texture

One of the sources of errors affecting the uncertainty of measurements of the surface texture is the measurement noise, called dynamic noise, which is a combination of the internal noise of the measuring instrument and the influence of the external environment. The article presents the comparison of measurement noise for instruments using various methods of measuring the surface texture that were used during the implementation of the project No. PBS2/A6/20/2013//NCBiR/24/10/2013 “Research and evaluation of reliability of modern methods of surface topography measurements in micro and nano scale”.

In recent years, there has been a rapid development of new methods for measuring the geometric structure of the surface, enabling the assessment of surface stereometry in both the micro-and nanoscale. Many publications have been written on the subject of measurement methods and evaluation [1][2][3][4]. The most commonly used methods of measuring surface stereometry are contact profilometry with a mapping blade, profilometry using confocal heads, coherent scanning interferometry, confocal microscopy and atomic force microscopy. The obtained values of the surface geometrical structure parameters are influenced -regardless of the measurement method used -by many factors that are components of the measurement uncertainty. One of them is the measuring noise, which is a component of the output signal. Sources of its formation can be divided into those resulting from the principle of the device operation itself and from the impact of the external environment on the measurement process. The most important factors generating measuring noise include: instability of electronic systems, vibrations caused by mechanical drives, external vibrations at the place of the device installation, air movement caused by ventilation or air conditioning, and electromagnetic disturbances. Value and nature of the measurement noise depend on the design of the instrument and the measurement method used. One of the methods of assessing the measuring noise of instruments for measuring the geometrical structure of the surface is to measure the AFL type standard (flat interference plate) [5,6]. Parameters that characterize the measuring noise value can be the square root of the ordinates of the primary profile Pq or the roughness profile Rq: where: n -number of ordinates in the roughness profile, z -value of the ordinate of the roughness profile.
Surface measurements with low roughness are particularly sensitive to disturbing factors. It is therefore important to include noise value in the uncertainty budget of this type of measurement.

Measurement results
Measuring noise studies were carried out in several research centers in Poland. They were carried out on fourteen instruments for assessing the geometric structure of the surface, representing the first two classes of methods, in accordance with the classification system presented in PN-EN ISO 25178-6: 2011 [7], because they give the output in the form of a linear profile z(x) or in the form of spatial topography z(x, y). For the purposes of research, the instruments were divided into two groups: representing optical methods (symbol O) and contact methods (symbol S). The list of devices and their symbols is included in tab. I. O5 Profilometric -confocal head 6.
S8 Contact profilometry -induction head During measurements of the geometric structure of the surface with the use of profilometric instruments using the AFL standard, due to the long time of these measurements, a clear impact of cyclic changes in ambient temperature on the results obtained was also observed, which was manifested by the occurrence of corrugation in the direction perpendicular to the axis of measurement [8]. This effect can be weakened by the use of thermal shields [9].

Measurement of AFL type standard on atomic force microscope
In order to check the geometrical structure of the AFL type standard, it was measured by means of an atomic force microscope using a SCANASYST-AIR type microbe (cantilever) ( fig. 1). Basic dimensions of the microprobe blade according to the manufacturer's data are as follows: blade height 2.5÷8 μm, front angle 15±2.5°, rear angle 25±2.5°, nominal radius 2 nm.    3 shows the ordinate distribution, the Abbott-Firestone curve, and the directional structure of the surface of the pattern. Five hundred and twelve profiles were extracted from the measured surface, for which the average values of basic parameters of the primary profile and the roughness profile were calculated using the Gauss filter λc = 0.025 mm (tab. III). An example of a primary profile is shown in fig. 4, and its frequency spectrum in fig. 5. Based on the analysis of the results, it can be concluded that geometrical structure of the AFL pattern is random, isotropic, with inequalities with an amplitude of several nanometers.

Investigation of the influence of the mapping blade feedrate on the measuring noise value
An important parameter that should be selected before measuring with a contact profilometer is the speed of the mapping blade. In order to determine the influence of the tip feedrate of the measuring tip on the value of the measuring noise, measurements of the AFL type standard were performed. For this purpose, a contact profilometer with an interferometric head was used, equipped with a measuring tip in the form of a 2 μm diamond needle with a 90° cone angle. The sampling interval was ΔX = 0.125 μm. Measurements were made for the following speed of the mapping blade v: 0.1; 0.25; 0.5; 1; 2 mm/s. Fig. 6 shows the obtained roughness profiles for selected mapping blade feedrates. Tab. IV, however, contains values of Pq and Rq parameters (after applying a Gaussian filter λc = 0.08 mm) depending on the feedrate of the mapping blade.  Based on the results obtained, it can be stated that up to the feedrate v = 0.5 mm/s, the value of measuring noise remains constant, while for velocity v = 2 mm/s, it is almost four times higher.

Comparative measurements of measuring noise
Comparative measurements of measuring noise were carried out for six optical instruments and eight contact instruments. The measuring section with a length of 1.25 mm was analyzed. Roughness profiles after the application of the Gaussian filter λc = 0.08 mm, obtained for optical and contact instruments respectively, are shown in figs. 7 and 8.  Optical instruments O1, O2, O3 and O5 as well as the S1 contact device are profilometric devices, in which the feedrate is carried out by the XY table. In their case, the highest measuring noise values were obtained. The distribution of the ordinates of the measured profile on the AFL standard is usually similar to the normal distribution. The exception is even distribution profile obtained with the S6 contact device.

Summary
The study showed that of all instruments, the O4 optical device using the method of coherent correlation interferometry is characterized by the lowest measuring noise value. even smaller than when using the atomic force microscope. This may be due to the fact that the sampling distance for this instrument was ΔX = 1.66 μm, i.e. it was more than ten times larger than for the atomic force microscope. Among the contact devices, the smallest values of measuring noise are characteristic for S2 and S3 devices, equipped with interferometric heads. The highest measuring noise values are characterized by optical instruments O1, O2, O3 and O5, equipped with confocal heads, and an S1 contact device with an induction head. The high noise figure for these instruments is due to the fact that the horizontal axis feedrate is realized by the XY measuring table.
For the O6 instrument, i.e. the confocal laser microscope, periodic disturbances can be observed on the roughness profile, resulting from errors in joining individual measurements due to the need for using the stitching function suitable for optical instruments. Conducted tests confirmed that for many instruments, measuring noise is a significant component of the uncertainty budget of the measurements of the surface geometric structure. For this reason, it is advisable to check noise to ensure reliable measurements.