*2.2. Production of 3D-Microtexture through Laser Ablation*

For the production of the desired cube-shaped 3D microtexture, a laser work station (microSTRUCTvario, 3D-MICROMAC, Chemnitz, Germany) in combination with a femtosecond laser (SPIRIT, Spectra Physics, Rankweil, Austria) were used. The laser delivered 350 fs pulses at 200 kHz with an average output power of 4 W at a wavelength of 1040 nm and 1.6 W at 520 nm. The laser beam was linearly polarized and the polarization direction could be flipped by 90◦ during the ablation process. To focus the laser beam, a telecentric scanner optic (Linos F-Theta Ronar, QIOPTIQ Photonics GmbH & Co. KG, Feldkirchen, Germany) with a focal length of 100 mm providing a focus spot radius of 6 μm for the 520 nm wavelength was used. For all test samples, a scan speed of 1000 mm/sec and a hatch distance of 5 μm were used.

The disc specimens for the tribological tests were produced in two steps using an average laser power of 385 mW. Firstly, a pattern along the x direction consisting of stripes with a width of 40 μm was scanned using the aforementioned hatch distance of 5 μm. After 15 consecutive scans, the polarization was flipped by 90◦ (symbolized by the yellow cross in Figure 4a), and this procedure was repeated 11 times. The laser polarization of the final scan cycle was in the x direction and parallel to the scan direction (the red arrow in Figure 4a represents the scan direction and the yellow arrow represents the polarization of the final scan cycle). The bottom of level 3 of the first trench, as shown in Figure 4a, was formed by this first ablation cycle, and the depth with respect to level 1 was approximately 35 μm. Secondly, the same procedure was performed with scans along the y direction. The polarization was

flipped similarly as in the first step, and consequently, the final scan ended up with a polarization along the x direction (perpendicular to the scan in the y direction). This second step formed the 35 μm deep bottom level of Section 2 (shown in Figure 4a) and a deeper bottom level (level 4) at the intersection with level 3. Surprisingly, the surface quality of Section 2 was not equal to Section 3. The fact that the laser beam is astigmatic could be one of the reasons for such an outcome. To rule out the laser beam astigmatism in our experiments, it was decided to do only scans along the x direction but to keep the polarization flip procedure after 15 consecutive scans. By using this procedure, the trench width was increased from 40 to 200 μm and the interfering effects from the generated side walls of a too narrow trench were therefore limited. In addition, the number of repetitions was reduced from 11 to 7 and, as a consequence, the depth of the obtained trench was reduced from 35 to 25 μm respectively, as shown in Figure 4b. One can clearly see traces from the 5 μm hatch pattern (small lines) along which pinholes with diameters of approximately 2 to 3 μm were formed. Because the polarization of the final scan cycle was along the x direction, the generated LIPSSs are orientated in the y direction [23,24]. As a consequence, the hatch lines intersect the LIPSS, or at least modulate the depth of the LIPSS, which leads to the growth of sub-micrometer pinholes of approximately 2 to 4 μm in diameter. Such a growth mechanism in SiC was previously simulated and investigated, and showed that LIPSSs correlate well with the slot waveguide characteristics in high refractive index material [20]. It is believed that such a model may also be applied to metals because, according to theoretical and experimental research, deep grooves in LIPSSs behave as plasmonic slot waveguides [25]. Moreover, previously published models predict a field distribution for deep slots which can generate nanometer-sized pinholes or cross periodic structures respectively [25,26]. The model previously developed by the present authors describes the growth of such nanometer pinholes towards micrometer sizes at the locations of interruptions or distortions of LIPSSs [20]. In earlier work on SiC, a certain threshold for pinhole growth was identified and the laser power was consequently reduced from 385 to 238 mW, while all other conditions were kept constant [20]. There was a significant reduction in the pinhole formation rate but at the cost of a reduction of the trench depth from 25 to 18 μm, as shown in Figure 5a.

**Figure 4.** Micrographs: (**a**) 3D microtextured disc; (**b**) test trench (hatch intersect the LIPSS).

An increase of the number of scans in order to reach a depth of 25 μm would have required an unwanted longer production time; therefore, it was decided to test the approach for which the final scan cycle was a combination of a scan in the x direction with a polarization in the y direction. By using such a procedure, it is expected that the generated LIPSS would be nearly parallel to the hatch lines, the number of intersections among hatch and LIPSS would be lower and the formation and growth rate of the pinholes may be reduced.

**Figure 5.** Surface quality after laser ablation: (**a**) reduced laser power and polarization of final scan cycle parallel to scan direction; (**b**) increased laser power and number of scans, but polarization of the last scan cycle perpendicular to the scan direction.

Furthermore, the laser power was increased from 238 to 279 mW, while the number of scans per constant polarization cycle was reduced from 15 to 10; in return, the number of cycles increased from 7 to 14. The obtained results verified our hypothesis and expectations: the obtained surface quality was close to that of the previous test performed with a laser power of 238 mW along with the fact that a trench depth of 25 μm could be again obtained as previously, as shown in Figure 5b. As more than one parameter for this test was simultaneously changed, two supplementary tests were conducted in order to confirm that our concept of polarization flipping after a certain number of scans, combined with a final scan cycle having a laser polarization perpendicular to the scan direction, significantly contributes to a smooth surface quality after laser ablation. Using the same parameters and procedure as previously, but skipping the polarization flipping, it is possible to compare the results obtained from a laser polarization and a scan both parallel to the x direction to the results obtained from a laser polarization along the y direction and perpendicular to a scan in the x direction, as shown in Figure 6a,b. The obtained results obviously support the hypothesis that in metals (as in SiC), pinhole growth is linked to the number of interruptions of the LIPSS. After these experiments, it came to the authors' attention that the simple rule of orientating the laser polarization perpendicular to the scan direction is the most effective measure to suppress pinhole growth, from negligible nanometer size pores up to micrometer dimensions. However, additional polarization flipping still has its benefits, i.e., contributing to pinhole suppression, because practically, it is impossible to obtain undisturbed gratings like LIPSS patterns on a larger ablation area; eventually, some pinholes will grow over time. LIPSS formation and subsequent pinhole growth are favored either by long durations at low scan speeds or a high number of scans at higher scan speeds. If the polarization is flipped within such a time or scan interval, the existing LIPSSs are reorganized along a new direction, the already generated pinholes are then removed (providing that their size is still not too big) and the growth cycle has to start again. Therefore, it was decided to use a growth cycle reset after every 10 consecutive scans and, as depicted in Figure 7b where the final cycle was performed with a laser polarization perpendicular to the scan direction, it may be observed that a further reduction in pinhole density with respect to the results shown in Figure 6b (same laser parameter but at constant polarization direction perpendicular to the scan direction) may be obtained.

**Figure 6.** Surface quality after laser ablation without any polarization flipping: (**a**) polarization permanently parallel to scan direction; (**b**) polarization permanently perpendicular to scan direction.

**Figure 7.** Micrographs: (**a**) LIPSS intersections perpendicular to their orientation where small pinhole formation starts; (**b**) same laser procedure as in Figure 5b but with enlarged scale (polarization of the last scan cycle was perpendicular to the scan direction).

However, the aforementioned simple rule of orientating the laser polarization perpendicular to the scan direction has one main disadvantage which becomes more important as a trench gets deeper after several scans. Due to Brewster angle effects and a resulting higher ablation rate in the direction of the laser polarization, small scratches in the trench wall grow faster and end up in a distinct score pattern [21]. After several consecutive scans, the pattern in the trench wall initiates a corresponding structure formation on the LIPSS-covered trench bottom, as shown in Figure 7a. Small pores are arranged in strings which are parallel to the polarization and perpendicular to the scan direction. The LIPSS are intersected perpendicular to their orientation, and after a number of several consecutive scans, the pores grow together to form bigger identities [21]. Any interruption of the LIPSS (for example a small bump) can trigger the formation and growth of pinholes, as depicted in Figure 7a. The front and end sides of the surface bump represent an interruption of the LIPSS (encircled areas in Figure 7a), and the growth cycle of pinholes predominately starts at this location. LIPSS striking tangentially the surface bump are not interrupted and a lower degree of pinhole formation was observed. The observations mentioned above (formation and growth of pinholes due to LIPSS intersections) are also interrupted by polarization flipping and have to start again from scratch, thus contributing to smoother surface quality and reduced pinhole formation. It is worth noting that for a better visualization of the aforementioned observations shown in Figure 7a, a laser with a wavelength of 1040 nm was used in

order to obtain larger LIPSSs. Moreover, larger and deeper LIPSSs may provide further advantages when they are tribologically tested under oil-lubricated conditions. In the current work, 3D textures without any LIPSSs were used as the top contact area. It is believed that a further improvement of the current 3D microtexture may be achieved through a two-step procedure: firstly, the laser ablation of the trenches along both the x and y directions should be performed using a laser with a wavelength of 520 nm (which showed a lower tendency for pinhole formation in comparison to the laser with a wavelength of 1040 nm), and further applying polarization flipping under the condition that the final scan cycle should be made with a perpendicular polarization with respect to the scan direction; secondly, after producing the 3D microtexture, the whole surface area should be exposed to the laser with a wavelength of 1040 nm in order to cover the microtexture with a LIPPS-based nanotexture, such as that shown in Figure 7a. It is believed that the current tribological results under oil-lubricated conditions could be improved by such a hierarchical micro/nanotextured surface in future tribological investigations. Furthermore, it is believed that the adhesive strength and the wear resistance of the antifriction coating could be improved through this novel manufacturing procedure.

#### *2.3. Evaluation of Dynamic Friction Coe*ffi*cients*

Stepwise load-varying tribological tests were performed using a SRV-tribometer (SRV®4, Optimol Instruments Prüftechnik GmbH, Munich, Germany). In these tests, friction coefficients were measured for short-time (5 min) oscillating tests at several different normal load values ranging from 25 to 200 N with steps of 25 N (as shown in Figure 8) under different lubrication conditions for as-received cylinders against different discs surface states: benchmark blank, antifriction coated benchmark, 3D microtextured or 3D microtextured/antifriction coated, as shown by the test matrix listed in Table 2. Furthermore, a supplementary test with a coated microtextured specimen, with the 3D microtexture oriented at an angle of 45◦ (instead of 90◦) to the direction of motion of the cylinder, was also performed in order to investigate any possible influences of the microtexture orientation on its tribological behavior. Table 3 lists some of the important test parameters used for the determination of dynamic friction coefficients. For the determination of dynamic friction coefficients, the tribometer was programmed to perform high-frequency friction signal acquisition (FSA) at the beginning and end of each load level for duration of 0.2 s at a sampling rate of 1 kHz. This FSA signal analysis enables a highly precise time-resolved analysis of the friction force and coefficients of friction, which are otherwise not possible using the normal signal acquisition parameters of the SRV-tribometer. The obtained FSA raw signals were then postprocessed in order to calculate averages of dynamic coefficients of friction. This postanalysis consisted of eliminating 10% of the raw signal at the beginning and at the end of a half-cycle, i.e., keeping only 80% of the raw signals for each half-cycle and averaging these values over approximately 10 half-cycles, as shown by the light green zones in Figure 9. From Figure 9, it may also be observed that the first peak present at the onset of movement (in either direction) is due to the adhesive part of the friction, also known as static friction or stiction.

**Figure 8.** Schematic representation of the load-varying tribological tests used for the determination of the dynamic friction coefficients.


**Table 2.** Test matrix of all tribological investigations performed in the present study.

\* Commercially available high-quality multi-grade hydraulic oil (ISO VG37).

**Table 3.** Main parameters used for the tribological investigations.


**Figure 9.** Example of raw signals obtained using FSA (high frequency signal acquisition) used for the calculation of both static and dynamic coefficients of friction (example shown: (**a**) benchmark (untextured samples) with oil at *F*<sup>N</sup> = 175 N and (**b**) 3D microtexturing with oil at *F*<sup>N</sup> = 175 N).

#### *2.4. Evaluation of Wear Resistance and Long-Term Friction Behaviour*

The wear resistance of the coated specimens (benchmark, 3D microtextured and 3D microtextured at 45◦) also represents a major interest prior to their deployment in real industrial applications. The reason that only the coated specimens were selected for the investigations of wear resistance was based on the fact that:

1. oil-lubricated conditions may possess certain drawbacks for some specific highly technical industrial applications; thus, antifriction coated specimens were selected, since they are widely used in industrial applications.

2. it was considered useless to investigate the wear of uncoated specimens under unlubricated conditions, since these conditions are usually never used in the industry.

The wear resistance of the previously specified antifriction coated samples was determined through long duration tribological tests (120 min). Except for the total test duration (120 min at 1 load level instead of 10 load steps of 5 min each), all other test parameters were kept identical to those used for the determination of friction coefficients, as listed in Table 3. Due to the restricted number of available samples, these long-term tests were performed at a normal load value of 125 N only, i.e., at a load value high enough to produce a measurable wear on the investigated samples.

The overall wear of the tested samples (cylinders and discs) was determined using the aforementioned laser optical surface analysis apparatus (Keyence VK-X250/260). For both cylinders and discs, a direct measurement of the wear scar volume was performed using a height threshold-based volume measurement. It is worth noting here that for both antifriction coated 3D microtextured disc samples (90◦ and 45◦), the measured wear volume values also account for the small proportion due to the valleys of the 3D-texture present at the bottom of the wear scar, and thus, the measured and presented values are slightly higher than the real wear volume values.

#### **3. Results and Discussions**

#### *3.1. Characterization of Benchmark, Coated, 3D Microtextured and 3D Microtextured*/*Coated Samples*

For the blank uncoated benchmark samples (cylinders and discs), the only characterization performed was the measurements of their surface roughness, which were already mentioned in Section 2.1 and are additionally listed in Table 4. For the untextured antifriction coated specimens (coated benchmark), the application of the aforementioned antifriction coating to their surfaces increased their surface roughness significantly, as shown in Table 4.

**Table 4.** Surface roughness values of untextured uncoated (blank benchmark) and untextured antifriction coated (coated benchmark) disc specimens.


A detailed measurement and analysis of the microtextured disc specimens was performed in order to ascertain the accordance of the dimensions of the produced microtextures with their initial desired dimensions. Figure 10 shows a detailed typical topographical analysis of two selected 3D microtextured disc samples; their measured specific dimensions are listed in Table 5. These dimensions correspond quite exactly to the initial desired dimensions presented in Section 2.1, thus showing that the production of 3D-microtextures using an ultra-short pulsed laser may be performed with a good level of accuracy. The ratio of the nominal microtextured area of the top plateau in comparison to the benchmark (benchmark = 100%) was calculated using the measured side dimensions and periods listed in Table 5, resulting in a value of 20.9%, which corresponds quite adequately to the desired ratio (20%) calculated from the desired microtexture dimensions presented in Section 2.1. These results show clearly that the production of 3D microtextures on steel may be achieved with a very high level of precision.

**Figure 10.** Typical topographies of two selected 3D surface microtextures produced on disc specimens: (**a**) first sample; (**b**) second sample.


**Table 5.** Typical dimensions of produced 3D microtextures on disc specimens.

As mentioned earlier, some of the 3D microtextured samples were then coated with an antifriction coating (properties shortly listed in Table 1). Typical topographies of such 3D microtextured and antifriction coated samples are shown in Figure 11. The first obvious observation that could be made from a comparison between Figures 10 and 11 is that the produced surface microtextures are no longer quadratic, and are not as sharply defined as previously: actually, the microtextures are in some way rounded by the presence of the antifriction coating. Furthermore, it may be easily observed that the valleys were partially filled with the antifriction coating, resulting in a reduction in the total height difference between the different plateaus of the 3D-texture. From the typical topographies presented in Figures 10 and 11, it may be easily concluded that the coating process used in the present study to apply the antifriction coating strongly modified the surface topography of the microtextured samples. The height of each plateau of the microtextured and coated samples was measured using the same procedure as for the microtextured samples. However, the lateral dimensions (bx, by) and the periods (Px, Py) of the different plateaus were not determined due to their undefined rounded profiles, which are difficult to measure. The average height differences from 3 different measurements between each plateau are listed in Table 6, and show that the height differences between the plateaus had decreased due to a partial filling of the microtexture valleys by the antifriction coating, as mentioned previously.

**Figure 11.** Typical topographies of two selected 3D microtextured and antifriction coated disc specimens: (**a**) first sample; (**b**) second sample.


#### *3.2. Evaluation of Dynamic Friction Coe*ffi*cients*

The results obtained from the load-varying tests performed on all sample configurations under study are presented in Figure 12, in which each point represents the calculated average of the friction coefficient (calculated from the sliding section of the obtained curve as shown in Figure 9) for the specified applied normal force. For the benchmark sample under oil-lubricated conditions (Figure 12a, black curve), it may be observed that the sliding coefficient of friction is very stable throughout the load range under study (which is shown by the relatively small error bars), and that its value remains at around 0.10. For the coated benchmark sample (Figure 12a, red curve) under dry conditions, the coefficient of friction becomes very unstable (large error bars) and is higher than the value of the benchmark under oil-lubricated conditions. Furthermore, its value stays at around 0.22 up to a load of 100 N and then increases abruptly to a higher value of approximately 0.30 for higher loads. This increase of the friction coefficient has been visually correlated to the coating failure during the test.

**Figure 12.** Sliding coefficients of friction as a function of applied load: (**a**) benchmark with oil or antifriction coating AF320E, (**b**) 3D microtexture (90◦) with oil or antifriction coating AF320E, and (**c**) comparison of all 3D microtexture results (note: different scales for y-axis).

For the 3D microtextured samples (Figure 12b), both the sample under oil-lubricated conditions (black curve) and that with an antifriction coating (red curve) show relatively stable coefficients of friction (small error bars) throughout the load range under study. For the 3D microtexture under oil-lubricated conditions, the coefficient of friction remains around 0.14 and for the 3D microtexture with an antifriction coating, friction coefficient stays around 0.20. By comparing the results of Figure 12a,b, it may be observed that under oil-lubricated conditions (black curves), the 3D microtextured samples exhibit slightly higher coefficients of friction then the benchmark (COFbenchmark ~0.10; COFmicrotexture ~0.14), and that the friction stability is similar for both samples, showing that the use of 3D microtextures does not bring any significant benefits under oil-lubricated conditions. However, with the use of an antifriction coating (red curves of Figure 12a,b), one may observe that the stability of the coefficients of friction is drastically increased (comparison of error bars), and that the coefficients of friction are significantly smaller for the 3D microtextured samples than for the benchmark samples, showing that the use of 3D microtextures along with an antifriction coating under unlubricated conditions represents an interesting alternative to oil-lubricated, untextured surfaces (benchmark).

It is believed that for the benchmark samples, a depletion of the antifriction coating occurs during the tribological tests, and that the coating is actually pushed out of the contact zone, thus exposing the underlying blank steel of the disc specimen to the cylinder, resulting in high and unstable coefficients of friction, especially for high load values. For the tests with the 3D microtextured and coated samples, a depletion of the antifriction coating also occurs, but only on the first microtexture level; the underlying levels of the 3D microtexture itself serve as lubricant reservoirs in which particles of the depleted antifriction coating from the top microtexture level are stored and continuously smeared back on the top level of the microtexture, thus enabling a kind of continuous lubrication regime which results in low and stable friction coefficients for all load values under study.

The results from tests performed on 3D microtextured samples with an antifriction coating but with the microtexture aligned at an angle of 45◦ from the sliding direction (instead of 90◦ for the previously presented results) are presented in Figure 12c (results of Figure 12b are incorporated in Figure 12c in order to compare easily the effect of the microtexture angle). From Figure 12c, it may be easily observed that the modification of the microtexture angle (from 90◦ to 45◦) with respect to the sliding direction tends to decrease the coefficient of friction towards the value of the microtextured samples at an angle of 90◦ tested under oil-lubricated conditions for low load values only (<125 N), and for higher loads, the coefficient of friction of the 45◦-oriented microtexture increases back to approximately the same values as for the 90◦-aligned microtextured sample with an antifriction coating. However, these observations should be taken with caution, since error bars of the results obtained may indicate that this behavior is not as obvious as observed, and that these observations may be solely due to the intrinsic scattering nature of the measured coefficients of friction. More detailed investigations are necessary in order to assess the veracity of the previous observations. Nevertheless, by considering the obtained error bars, one may conclude without any doubt that the use of a 45◦-oriented 3D microtexture (instead of a 90◦-aligned 3D microtexture) with an antifriction coating tends to slightly lower the coefficients of friction, but that the coefficients of friction never fall below the values obtained for the 3D microtextured samples tested under oil-lubricated conditions.
