Dry Friction Between Laser-Patterned Surfaces: Role of Alignment, Structural Wavelength and Surface Chemistry
- First Online:
- Cite this article as:
- Gachot, C., Rosenkranz, A., Reinert, L. et al. Tribol Lett (2013) 49: 193. doi:10.1007/s11249-012-0057-y
- 579 Views
The ability to tune friction by tailoring surface topographies at micron length scales and by changing the relative orientation of crystallites at the atomic scale is well established. Here, we investigate if the two concepts combine, i.e. if the relative orientation of surfaces affects dry friction between laser-textured surfaces. Laser patterning was used on austenitic stainless steel substrates and on tribometer testing balls made of 100Cr6 to create linear periodic arrays with different structural wavelengths or periodicities (5, 9 and 18 μm). Pairing each substrate with a ball of the same periodicity, the different arrays were subjected to dry sliding tests at 0°/90° relative alignment between the linear patters. We observe that the patterning reduces friction after running-in. The reduction increases with decreasing wavelength and also depends sensitively on the relative alignment and the chemistry of the sliding surfaces. Our results highlight the possibility to create tailored contacting surface geometries leading to tunable frictional properties.
KeywordsLaser interference patterningDry frictionInterlocking
Most machines contain components with loaded solid surfaces that rub together. The interaction of such surfaces produces friction and also results in mechanical damage to them. Thus, tribological phenomena play a decisive role in diverse systems. For many years, researchers have sought to alleviate these problems and to understand their origin. The early work of Bowden and Tabor  showed that one influential aspect on friction is the interaction of asperities. They can be welded together and released by applying a critical shear stress τc. However, studies at different spatial scales revealed that tremendously small shear stresses between atomically flat surfaces are possible when their surface corrugation does not match . Such a state of nearly vanishing friction was introduced by the pioneering work of Hirano and Shinjo  for incommensurate, i.e. non-matching contacts, and confirmed experimentally by Dienwiebel et al.  on the nanometre scale. Recently, Sondhauß et al. have demonstrated the influence of a mesoscopically patterned silicon surface on the frictional response. They used a silicon substrate which was patterned using a focused ion beam microscope (FIB) and, as a counterbody, titanium spheres (without pattern) with different radii attached to a slightly modified AFM cantilever. It was shown that the effective coefficient of friction has characteristic maxima depending on the groove widths in the silicon substrate and the corresponding radius of the titanium spheres .
The ability to control and modify frictional forces on different scales bears promise for the design of miniaturised systems, positioning devices and bearings . There are two major methods to manipulate friction: modification of the surface materials or texturing of the relevant surfaces [6, 7]. In particular, laser surface texturing (LST), introduced by the Etzion group for the texturing of mechanical seals, proved to be a promising technique for dry and lubricated contacts [8–11].
One possible approach to LST is laser interference metallurgy. It is based on the interference of laser beams from a solid-state laser, which creates well-defined surface topographies on the micron scale [12–15]. Moreover, due to a periodic laser intensity distribution and significant heating/cooling rates of up to 1010 K/s , metallurgical effects like melting, resolidification and the formation of intermetallic phases can be induced [17, 18]. In the present work, we apply the aforementioned technique to pattern both interacting surfaces, rather than just one, to control the involved contact geometries under dry sliding conditions. Using two-beam interference, linear patterns with varying periodicity of the laser intensity maxima were fabricated. Furthermore, the as-patterned contacting bodies were scrutinised with regard to their ability to interlock depending on the structural wavelength and the relative alignment. In addition, microstructural (X-ray diffraction) and chemical analysis (Raman spectroscopy) were applied to study the laser-treated specimens. Finally, experimental results of the frictional response will be presented and discussed.
2 Experimental Procedure
The intensity of each beam can be controlled using suitable beam splitters. The laser fluence was set to 400 mJ/cm2 for all specimens. Due to the absorption characteristics of the material used, the third harmonic at 355 nm was selected. The periodicity (line-spacing) was set to be about 5, 9 and 18 μm. All samples were irradiated at normal atmospheric conditions in air using one single laser pulse. Further details of the experimental setup have already been published elsewhere .
The topography was measured by a Zygo New View 100 white light interferometer (WLI) equipped with a 3-D imaging surface structure analyser.
The surface morphology was imaged by scanning electron microscopy (SEM) (FEI, Strata DB 235). In addition, a FIB microscope (FEI, Strata DB 235) was selected to prepare cross sections in order to investigate the microstructure.
XRD phase analysis measurements using grazing incidence geometry (GI-XRD) were carried out in order to obtain statistically relevant phase data of a significantly large specimen volume. For this purpose, a seven-axe diffractometer (PANalytical X’PERT MPD Pro) with a copper X-ray tube was used.
Raman spectroscopy was performed with a LabRAM ARAMIS instrument from HORIBA using a 532 nm laser beam without filter. The Raman microscope uses a back-scattering geometry, where the incident beam is linearly polarised and the spectral detection is unpolarised. The optical elements used in this analysis allowed a laser spot size of 5 μm (full width at half maximum) and depth penetration of ~1 μm, while the spectra precision was around 5 cm−1.
The tribological tests were performed with a nanotribometer using a ball on disc configuration in a linear reciprocating sliding mode (CSM Instruments) with a stroke length of 0.6 mm. The tribometer is based on a stiff cantilever, which acts as a frictionless force transducer in both vertical and horizontal directions. The static partner is loaded onto the substrate with a precisely known force using piezo-actuation. The normal load and the friction force are determined during the experiment by measuring the deflection of this elastic arm in both horizontal and vertical planes with two high-precision fibre-optical displacement sensors. Through a feedback loop, the piezo-actuation maintains the normal load independent of any surface irregularities. The normal force was set to 1 mN and the linear sliding speed to 1 mm/s. The deviation of the aforementioned normal load was less than 10 % with respect to the reference value in our experiments. The counterbody consisted of a typical 100Cr6 bearing steel ball with a diameter of 3 mm. Temperature and relative humidity were kept constant at 20 ± 2 °C and 45 ± 5 %.
3 Results and Discussion
3.1 Topographical Analysis
For a geometrical interlocking of the two patterned surfaces, the groove and asperity widths of the contact surfaces are essential. Figure 3 shows cross section plots of a patterned steel substrate and tribometer ball measured by WLI. The laser periodicity in both cases was set to 18 μm with a depth of around 1 μm. The analysed groove widths of the profiles presented in Fig. 3 range from 12 to 14 μm in the upper part and from 7 to 9 μm in the lower part of the structured substrate. The asperity widths for both surfaces range between 3 and 10 μm. Therefore, an interlocking of the structures based upon the topographical analysis is generally possible.
Additionally, the line-patterns with varying periodicity were studied with regard to surface profile parameters and compared to each other.
Topographical parameters of laser-patterned samples in comparison with the unpatterned reference recorded by WL
34 ± 5
26.32 ± 4
94.45 ± 11
−1.5 ± 0.2
3.12 ± 0.8
326 ± 12
626.44 ± 57
860 ± 49
0.84 ± 0.03
3.59 ± 0.26
305 ± 6
484.89 ± 36
775 ± 11
0.52 ± 0.07
2.38 ± 0.19
354 ± 34
218.05 ± 15
1164 ± 107
1.45 ± 0.13
5.30 ± 0.53
Apart from the reference, the 9 μm pattern has the lowest positive Rskψvalue. A comparison with the corresponding cross section profile plot reveals a very homogeneous laser-pattern. During the laser-induced melting process, the material is shifted away from the intensity maximum towards the minimum positions. In case of larger wavelengths, the distance between the maximum and minimum positions is expanded and so the distance in between cannot be bridged by the molten material. As a consequence, additional topographical maxima may appear with different heights.
The slope RMS, SRMS, is listed for the various specimens. This parameter is particularly interesting because it represents the square-root of the second order spectral moment, which can be found in literature as (m2)1/2, and is roughly inversely proportional to the real area of contact . As expected for similar structure depths, SRMSψ is the highest for the smaller structural wavelength.
3.2 Microstructural and Mechanical Characterisation
Even in the positions of maximum laser intensity, the grain structure is not affected in an apparent way. Austenitic stainless steels are generally composed of the metastable γ-phase . Thermal simulations were performed for the different structural wavelengths which have not shown any strong difference concerning the achievable temperatures in the positions of maximum and minimum intensity or the respective heating/cooling rates. Accordingly, the maximum temperature was approximately 1,900 K (in the maximum position) which is clearly above the corresponding melting range of the austenitic steel used (Tm is between 1,693 and 1,747 K ).
3.3 Surface Chemistry
It is worth noting the change of the relative intensities of the peaks located at around 590 and 640 cm−1. As expected, similar intensities with respect to the background are observed in the non-irradiated and minimum irradiated samples, whereas the intensity of the peak located at ~640 cm−1 increased in the radiation maxima zones. Since the exact vibration of this mode is not clear in the literature [25–31], the interpretation of this observation is not straightforward. Besides FeO, this mode may have contributions from magnetite (Fe3O4), maghemite (γ-Fe2O3) and hematite (α-Fe2O3), due to phase transformations induced by high-temperature processes [30, 31].
Moreover, as shown in Fig. 10, the peaks at ~270 and ~640 cm−1 are shifted between the spectra of the 99.99 % pure iron and the radiation maxima zone, and the intensity/background ratio of the mode associated with (Fe, Cr)O3 species decreased in the maximum radiation zone. The synergy of these evidences strongly indicates a modification of the oxidation state of the surface after laser patterning.
3.4 Tribological Results
A comparison of Fig. 11c and d further implies a strong effect of the different structural wavelengths.
It is known from the literature that the steel–steel friction coefficient under dry and clean conditions is smaller in the presence of these oxide layers (typically 0.27) than in their absence (μ ~0.80) . The friction coefficient of the reference in the beginning is in good agreement with the reported μ of 0.27.
The initial values of μ of the laser-patterned surfaces are approximately 30 % higher compared to the reference surface for a 0° orientation. This may be attributed to an interlocking effect as well as a modified chemical and microstructural state of the laser irradiated samples in which the composition of the oxide layers differs from the pristine material surface and the defect density (e.g. dislocations) may be increased. Subsequent to the initial situation, μ of the laser-patterned surfaces decreases independent of the used periodicity. An explanation for this decline after some run-in could be that the induced patterns gradually degrade, which can reduce the interlocking of asperities. The local properties of the patterned specimens become similar to those of the reference.
After the sudden drop of μ, an increase can be observed which may be linked to further flattening of the laser structures causing a larger real contact area and a changed surface chemistry.
Furthermore, after a sliding time of around 90 s, a deviating slope for all the specimens is visible. During sliding, oxide wear particles are created and shifted to the reversal points of the slider thus accumulating at the end of the wear track, as indicated by light microscopy. As a result, the friction regime is changed because the oxide layer is gradually removed and the friction coefficient approaches the reported value of 0.8 for the steel–steel contact without oxide layer. While the surface chemistry and surface topography certainly play important roles concerning the initial values and various slopes in the friction curves, they cannot explain the significant differences in μ for 0° and 90° orientation. One difference between the different orientations is their relative contact areas. They would yield different total friction assuming local friction-load relationships of the form Fs = σ0Areal + αpL. Fs and L denote the friction force and the normal load on a representative surface element, respectively, whereas the system-dependent σ0 and αp stand for an offset and a proportionality coefficient . This constitutive relation holds all the way down to the nanometre scale in the presence of lubricants or contaminants, with values in the order of 0.1 leading to a reduced but not vanishing friction . We will explore the possibility of contact area being important in a separate study.
As far as the wavelength dependency is concerned, we suppose on the one hand that this experimental result may be traced back to the mean square slopes of the measured profiles. The mean square slope SRMS shows a maximum value for a structural wavelength of 5 μm whereas the smallest SRMS can be found for 18 μm. This value is in leading order inversely proportional to the real contact area . On the other hand, the observed friction reduction may also be a consequence of fabrication errors during the patterning process. The resulting laser patterns are not perfectly sinusoidal and thus the interlocking could be complicated for a 0° alignment and a decreasing structural wavelength.
In general, the tribological behaviour of austenitic stainless steels is strongly affected by the formation of strain-induced ε-martensite . According to Huebner et al. , the martensitic transformation might occur by shear deformation of the lattice. During sliding, the dislocation density increases and, as a consequence, so does the stacking fault density. These steels have a relatively low stacking fault energy of about 18 mJ/cm2 which is comparable to that of Ag .
As already discussed, the stacking fault energy has a significant impact on the mechanical behaviour . Like ε-martensite, the formed stacking faults are of hexagonal symmetry and thus finally pile up to form ε-martensite .
Remarkable work in this field has also been published by Rigney and Hirth. Rigney  reported about the occurrence of large plastic strain fields during sliding deformation resulting in a strong rise of dislocations which organise themselves in dislocation cell structures. Typical values for dislocation densities after sliding contact of ductile metals are in the range of 1015 1/m2 which is consistent with simulation results from Kuhlmann-Wilsdorf [39, 40]. X-ray diffraction could not prove the existence of martensite after laser irradiation. However, further experimental work is necessary to study the potential formation of martensite and the influence of dislocations and stacking faults on the friction behaviour of the steel used, before and after laser patterning as well as after the friction experiments by high-resolution techniques. In particular, TEM and XPS experiments are a current matter of subject for more detailed studies of the microstructural and depth-sensitive chemical situation.
In conclusion, we have found that the dry friction between two laser-structured solids depends not only on the wavelength of the structuring but also on the relative orientation between the patterns. Surface chemistry and mechanical properties, such as hardness and yield strength, can differ for different wavelengths and might be partly responsible for why different wavelengths produce different friction. However, considering their influence cannot explain the critical role of alignment: in each experiment, the friction for misaligned surfaces turned out smaller than for a perfectly oriented interface, given the same wavelengths in both experiments. One possibility for the origin of the dependence on alignment is that the contact area is alignment dependent. We explore this idea more in detail in a companion paper .