The Effect of Surface Preparation on Result Deviations During Roughness Measurements Using the Contact Method

Abstract

This paper presents the influence of the preparation of the inspected surface on the deviations obtained when measuring the basic roughness parameters using the contact method. The problems described in the article are often encountered in industry when measurements are performed in mass production conditions, where a short measurement time is extremely important. Presented are the measurement results of five samples made of 1.0503 steel after the grinding process, subjected to various methods of preparation for roughness measurement: sample immediately after treatment, sample after treatment and after evaporation of liquid part of the coolant, sample after blowing with compressed air, sample after blowing with compressed air and washing in isopropyl alcohol, and the sample after blowing with compressed air and washing with acetone. The measurement results obtained from the contact profilometer provide the basis for developing recommendations regarding the preparation of surfaces for inter-operational and final measurements in the production process.

1. Introduction

Surfaces—the outermost layer of any workpiece—are created after, i.e., machining. The texture of the surface affects many properties of machine elements, such as contact mechanics, sealing, friction, and wear [1]. Surface topography is formed during the last stages of the machining treatment. Some properties, such as material contact, sealing, friction, lubricant retention, and wear resistance, are related to surface topography. Taking the above into consideration, inter-operational control during machining is a must to obtain the desired form and values of surface irregularities. Characterization of these surface irregularities is obtained by a number of measurement methods that examine the surface from different points of view and provide different information. Typically, there are three measuring techniques that can be divided into two types according to the results: quantitative and visualization. There are contact measurements, optical measurements, and other technical solutions, including atomic force microscopy (AFM) and confocal laser scanning microscopy (CLSM, scatterometry) [2,3,4]. There are two main measuring methods in the industrial environment, contact and optical, with the undoubtedly dominant role of contact profilometry. Contact surface profilometers make physical contact with a measured specimen using a sensitive stylus, which is drawn across the component surface. This can register surface deformations and variations as fluctuations in friction and applied resistance. It is suitable for roughness, waviness, and form measurements and is impervious to optical interference and lubricating elements. In engineering, one of the problems is that environmental conditions can vary widely between the machine shop and the inspection room. This is where the stylus method scores over other methods. The fact that the stylus touches the surface can be an advantage in two ways. The first is that the stylus can push aside debris and penetrate films that would otherwise mask the geometry. The second is that the stylus can be used to act upon the surface, as in the case of the scanning microscopes, to initiate discharge, disrupt fields, and behave in a proactive way as required [5,6,7].

Surface preparation for roughness measurements is often an omitted issue both during an experiment and in everyday metrology practice. Trumpold and Frenzel stated [7], “a surface is clean when any residual contamination fails to affect the functionality of that surface”. In terms of roughness measurements, it may be changed to: any residual contamination fails to affect changes in the obtained results with assumed accuracy.

In [8], the author states: “From the point of view of engineering metrology, there is usually not much time to prepare a specimen workpiece prior to measurement. All that can be attempted in in-process measurement is a contact wiper on the surface or a brush for internal holes. Sometimes, compressed air is used to clear the surface of debris and oil films […]”. Practically, the most common approach to cleaning surfaces before roughness measurements is cleaning through the blowing of compressed air against the surface. The other common methods include cleaning with a dust-free cloth or a cloth soaked in isopropyl alcohol or acetone; less commonly, sampling can occur after bathing in a cleaning liquid. Guides usually give a brief description of how to prepare the specimen for the measurement, e.g., “Before performing a measurement it is important to remove any oil, grease, dust or mist from the surface […]” [9,10]. Many authors also include a short sentence such as “sample after cleaning/preparation” or even without any information about such preparation [11,12,13,14,15,16]. The absence of such information is not a big mistake (although it is good engineering practice to include it), but inaccurate or completely incorrect surface preparation for measurement results in erroneous measurement results.

2. Materials and Methods

The test methodology was based on the measurement of a Talysurf Taylor Hobson contact profilometer, with an internal datum and gauge equipped with an inductive transducer. This type of equipment is widely used both in industrial conditions as well as in research and scientific units throughout Poland. Five samples made of carbon toughening steel (1.0503) were prepared. The dimensions of each sample were 22 × 22 × 16 mm. This is a medium carbon steel designation assigned to a group of steels that have a carbon content between 0.42 and 0.50 wt.%. This type of steel is used in a variety of applications; they are particularly well suited for parts that require high wear resistance and strength, such as gears, shafts, and bearings. Thus, it is a very representative material due to its common use in industry. The sample surfaces were ground on a surface grinder at a speed of Vc = 26.4 m/s, with a final pass cutting depth of 0.005 mm. For the treatment, a new corundum grinding wheel (60 grit) was installed. Emulgol ES-12 emulsifying oil provided by Orlen oil, at a concentration of 5%, was used as a lubricating fluid. The samples were mounted on a magnetic table and machined at the same time, and after machining, they were prepared for measurement. Such an approach eliminates the potential differences between sample—tool—machine conditions. Individual samples were designated as S_W—wet surface, measurement immediately (30 s) after machining; S_D—dry surface, measurement after evaporation of the liquid fraction (1 h after end of treatment); S_CA—measurement after blowing with compressed air; S_CA + I—dry surface, measurement after blowing with compressed air and after bathing in 99.9% isopropyl alcohol; S_CA + A—dry surface, measurement after blowing with compressed air and bathing in 99% acetone. Both isopropyl and acetone methods were performed using a short bath (15 s) in a glass container; after that, both samples were gently dried using a dust-free cleaning cloth (68 g/m²) on the bottom and all side walls, with the upper surface left free to evaporate in the inspection room conditions (22 °C and humidity 55–60%) for 15 min. Twenty topography measurements were taken on each surface using a Form Talysurf profilometer. The instrument gauge was calibrated using a 60 mm stylus arm with a 2 µm radius and 90° conisphere diamond tip (shown in Figure 1) on an E1-type glass hemisphere standard made of silicon nitride. The nominal radius of the standard is 12.5 mm with RONt uncertainty equal to 0.10 µm. The sum of the largest profile peak height and the largest profile valley depth of the P-profile within the evaluation length was Pt = 0.09 μm for 80% of the gauge range.

Measurement sections of 6.1 mm on a 5 mm field were taken. The test field on sample S_D is shown in Figure 2. The influence of waviness on roughness for each surface was insignificant. For example, the parameter values for sample S_CA + I filtered with Gaussian 0.8 cut-off were Rz = 2.04 µm, Rt = 2.79 µm, and Ra = 0.243 µm. The unfiltered values were Rz = 2.06 µm, Rt = 2.78 µm, and Ra = 0.240 µm. Examples of the filtered profiles obtained from sample S_CA + I (waviness and roughness) are shown in Figure 3.

Taking the above into consideration, for the purpose of the analysis, the waviness was not filtered from the roughness, and the parameters from the original raw profiles were determined to minimize the impact of artificial computational factors as much as possible.

3. Results

On the basis of measurements, the following amplitude parameters were calculated (according to ISO 4287:1999, 4288, 3274, and later to ISO 21920-1:3 [18,19,20,21]):

Rz—maximum peak-to-valley height of the profile within a sampling length.

Rt—maximum peak-to-valley height of the profile in the evaluation length.

Ra—arithmetic mean of the absolute departures of the roughness profile from the mean line.

After all measurements were taken, the mean and extreme values were determined, as well as the deviation and standard error for each surface, which are summarized in

Table 1. Obtained and calculated results.

Sample	Parameter	Max	Min	Mean	Std Dev	Std Err
S_W	Rz	2.830	2.050	2.320	0.203	0.0454
	Rt	3.950	2.520	3.390	0.409	0.0915
	Ra	0.290	0.198	0.247	0.022	0.0049
S_D	Rz	2.780	2.280	2.510	0.162	0.0362
	Rt	4.250	2.730	3.400	0.363	0.0812
	Ra	0.303	0.225	0.260	0.019	0.0043
S_CA	Rz	2.980	2.290	2.610	0.198	0.0443
	Rt	4.900	2.760	3.790	0.569	0.1272
	Ra	0.315	0.249	0.287	0.022	0.0049
S_CA + I	Rz	2.430	1.920	2.160	0.135	0.0302
	Rt	3.630	2.360	2.980	0.322	0.0720
	Ra	0.265	0.203	0.229	0.016	0.0035
S_CA + A	Rz	3.110	2.120	2.500	0.224	0.0501
	Rt	4.810	2.980	3.550	0.435	0.0973
	Ra	0.288	0.225	0.257	0.018	0.0040

.

Examples of the obtained mean profiles are shown in Figure 4. The scale for the S_CA + I (fourth from top) sample was increased to 1 μm to better show the surface irregularities.

After compilation, the largest differences in the mean results of individual parameters between all samples are presented as follows:

-

For the Rz parameter, the largest difference between the surfaces is 0.45 µm;

-

For the Rt parameter, the largest difference between the surfaces is 0.81 µm;

-

For the Ra parameter, the largest difference between the surfaces is 0.058 µm.

All of the above differences occurred between the S_CA and S_CA + I samples (Table 1). Since the Ra parameter is an arithmetic mean value, it shows the least susceptibility to the outliers and, as such, will be the least relevant for analysis.

The value of the Rz (Figure 5) parameter almost twinned for samples S_D and S_CA + A (sample with evaporated coolant liquid phase and sample after blowing and bathing in acetone). The outliers are the already mentioned samples after blowing (S_CA) and after blowing and bathing in isopropyl alcohol (S_CA + I).

For the total profile height Rt (Figure 6 top), very similar values were obtained for the sample straight from the machine and for the sample after evaporation of the liquid part (S_W and S_D, respectively). Once again, extreme values were obtained for samples after blowing and bathing in isopropyl alcohol. The arithmetic mean deviation from the mean line inside the elementary section (Ra shown in Figure 6 bottom) confirms the differences in the extremes, but the differences are too small to clearly determine the nature of the differences. In further discussions, Ra will be omitted.

In order to better illustrate the spread of the results, in Figure 7, the average values are compared against the maximum and minimum values.

4. Discussion

For the wet sample (S_W), the average value of the highest profile height (parameter Rz) was closer to the lower limit of the range (0.27 μm versus 0.51 μm). The Max−Min difference (range width) was 0.78 μm. For the dry (S_D) surface, the average value of the Rz parameter was located near the middle of the range. The entire width of the range of values obtained was 0.50 μm, and the average was below the center of the range (0.02 μm). The surface after blowing with compressed air (S_CA) similarly showed a tendency of the mean to the middle of the range (0.025 μm below), with the width of the range being greater than that of sample S_D but less than that of sample S_W (0.50 < 0.69 < 0.78). The surface after blowing and bathing in isopropyl alcohol showed the lowest values of the Rz parameter. The range of values was close to the S_D sample (0.51 μm for S_CA + I with 0.50 μm S_D). The average value was closest to the center of the range (0.015 μm below the center). The width of the results obtained for the surface after bathing in acetone was 0.99 μm. This was the widest range among all the results. The average value of Rz = 2.50 μm was 0.115 μm below the center of the range (Figure 7 top). The total height of the profile of Rt for the entire measurement section of the S_W sample was 3.39 μm, with a range of variation of 1.43 μm. The mean was the only one above the middle of the range (0.155 μm). Sample S_D had a mean of 3.4 μm, with a range width of 1.52 μm. The mean was located 0.09 μm below the center of the range. The post-blow surface of S_CA had the widest range of variation, with a width equal to 2.14 μm. In turn, its average value of the Rt parameter was also located as the second closest value to the center of the range (below the center by 0.04 μm). The surface after blowing and bathing with isopropyl alcohol had the narrowest area of variation (1.27 μm), and the position of the mean was closest to the center of the range (below by 0.015 μm). The surface after bathing in acetone had the second widest range of variation of the Rt parameter (1.83 μm) and was significantly different from the other samples, and the mean shifted towards the minimum (relative to the center of the range, it was as much as 0.345 μm—Figure 7 bottom). This is due to the random occurrence of peaks along the measurement trace—an example is shown in Figure 8—which is caused by the reaction of acetone on the measuring tip or surface. After examining the surface under the microscope, no physical cause (dirt, abrasive residue, etc.) was found to introduce such a deviation.

The most common method of sample preparation for measurements—the use of compressed air—gives the widest range of results for Rt (spread over 2 μm) and quite a similar range to the wet surface for the Rz parameter. After observation under the microscope, it is clear that lubricant fluid residues are still present on the surface, but they are dispersed around the whole area. A comparison between wet and blown surfaces is shown in Figure 9. It should be mentioned that it is also possible, depending on the quality of the compressed air installation, that additional fractions of oil or water may be introduced to the measured surface.

The narrowest range of results for Rz was obtained for the dry sample, and the second for isopropyl-cleaned samples. For Rt (also for Ra = 0.062 μm), the narrowest range was observed for the S_CA + I (isopropyl cleaned) sample. This method gave the lowest values for all three roughness parameters. Taking the above into consideration, it can be concluded that this method of surface preparation will give the most accurate values of the measurements (narrow range of variability and lowest values). The biggest disadvantage is the time needed to prepare samples due to the isopropyl bath after air blowing and then drying. A view of the surface thus prepared is shown in Figure 10.

5. Conclusions

The best way to prepare surfaces for roughness measurements is to bathe steel samples in isopropyl alcohol after blowing compressed air and letting the cleaner evaporate in inspection room conditions. Measurements of acetone-bathed samples showed that this type of preparation is subject to too much uncertainty. The wide spread of the results, as well as the presence of random peaks along the measurement trace, disqualifies this method from being used for very accurate measurements. The authors did not find a clear cause of these random peaks in the S_CA + A sample. There is a hypothesis that after the acetone dries, there could possibly be micro-scale static discharge.

The most common approach to surface cleaning for roughness measurement—compressed blown air—gave very poor results, so in normal conditions, this type should be a preliminary process only before using more accurate methods.

This article does not fully cover such a broad topic as preparing samples for measurement; however, it gives indications on what to pay special attention to before taking surface roughness measurements. Further research should be conducted to standardize sample preparation procedures.