tribology. Nevertheless there is still lack of knowledge about fundamental ... parameters in order to capture the tribological efficiency of structured surfaces.
Available online at www.sciencedirect.com
ScienceDirect Procedia CIRP 12 (2013) 456 – 461
8th CIRP Conference on Intelligent Computation in Manufacturing Engineering
Functional characterization of structured surfaces for tribological applications A.A.G.Bruzzonea*, H.L.Costab a DIME - University of Genoa, via Opera Pia 15, Genoa, 16145, Italy Laboratório de Tribologia e Materiais, Universidade Federal de Uberlândia, Uberlândia, 38400902, Brazil * Corresponding author. Tel.: +0-39-0103532894; fax: +0-39-010317750. Tel.: E-mail address: Alessandro.Bruzzone@unige. b
Abstract Engineered surfaces obtained by modification of surface topography have important industrial applications, many involving tribology. Nevertheless there is still lack of knowledge about fundamental aspects involved in the improvement of surface behaviour. Recent technological developments now permit us to texture surfaces in a flexible way and to assess the tribological efficiency of different microtopologies. Moreover, advances in surface analysis techniques provide methodologies that do not limit the investigation to microgeometry but allow the functional characterization of surfaces. This paper proposes functional surface parameters in order to capture the tribological efficiency of structured surfaces. © 2013 The Authors. Published by Elsevier B.V. Open access under CC BY-NC-ND license. © 2012 The Authors. Published by Elsevier B.V. Selection and/or peer-review under responsibility of Professor Roberto Teti. Selection and peer review under responsibility of Professor Roberto Teti Keywords: Surface; Tribology; Texture.
1. Introduction Functional surfaces exploit important physical and chemical phenomena that take place on surfaces, especially at the micro- and nano scale, and have played a fundamental role in the development of many advanced fields, such as electronics, information technology, energy, optics, tribology, biology and biomimetics [1]. In the field of tribology, estimates indicate large economic losses associated with friction and wear, as much as 5% of the GDP of developed countries [2], or 30% of world energy consumption [3]. Tribological applications of functional surfaces have mostly relied on the controlled modification of surface topography, microtexturing, as a means of improving lubrication and reducing friction and wear. Three main effects are expected to occur to improve the tribological performance of textured surfaces: 1) the entrapping of wear debris inside the texture pockets, avoiding debris presence between the two surfaces, which could cause abrasive or adhesive wear [4, 5]; 2) the existence of pockets of lubricant particularly useful as a secondary
source of lubricant [1, 6]; 3) the enhancement of a hydrodynamic pressure between the surfaces due to the presence of converging wedges in the surface [1]. Various authors have studied the tribological behaviour of textured surfaces, especially in recent years. These studies include numerical simulations and experimental evaluation of textured surfaces [4-8]. However, the effects of the different variables of the texturing on the tribological behaviour of textured surfaces are still not very well understood. To answer this question properly, it is necessary to identify precisely how each of the variables that define the pattern generated by the texturing process affects different aspects of the tribological performance. The characterization of surface microgeometry normally relies on the use of statistical and geometrical parameters [9, 10, 11]. The conventional techniques are often inadequate for specific industrial cases; industrial requirements relate to the surface’s function and depend on the physical phenomena occurring on the surface and its environment. Recent standards (ISO 25178 and 16610) address this drawback by shifting the focus on the 3D surface texture.
2212-8271 © 2013 The Authors. Published by Elsevier B.V. Open access under CC BY-NC-ND license. Selection and peer review under responsibility of Professor Roberto Teti doi:10.1016/j.procir.2013.09.078
457
A.A.G.Bruzzone and H.L.Costa / Procedia CIRP 12 (2013) 456 – 461
Fig. 1. Scheme of the experimental approach
Figure 1 shows the approach used in this study. The development of functional parameters includes the characterization of the tribological phenomena, by analysing the relationship between tribological performance and surface, load, lubricant (f1); the microgeometry of surface is described by texture parameters (f2). The relationship between texture parameters and tribological performance (f3) is examined in order to determine the most promising surface microgeometry to achieve the required functional, i.e. tribological, behaviour. In order to establish the relationships between global tribological parameters describing the macroscopic aspects involved in the phenomenon, and the specific surface parameters, this approach must face two problems: tribological phenomena 1) are nonstationary, and 2) occur in a very small dimensional scale. It is very difficult to measure local parameters such as pressure field, lubricant velocity, tension and deformation tensor fields within the lubricant and the surfaces in a microscale, and to relate parameters to texture features (areal, lines, points), whose selection is critical for assuring the required function. The experimental investigation reported in this paper assesses the suitability of functional surface parameters to capture tribological behaviour between metallic surfaces under full film lubrication.
The hydrodynamic effect between sliding surfaces occurs due the presence of a wedge-shaped film of fluid. The variation of lubricant pressure in a wedge is described by the Reynolds equation, which can model the hydrodynamic action on the basis of a convergent clearance space [2]. For structured surfaces, the microtopographies expected to offer the best tribological results consist of small cavities uniformly distributed over the sliding surface. Considering the small converging wedges generated by each one of these pockets, textured surfaces composed of a plurality of pockets could then be regarded as a set of microbearings. 3. Experimental Procedure AISI 01 GFS steel samples with dimensions 35 x 35 x 2 mm3 were cut from gauge plates with ground surfaces, in order to ensure parallelism of the surfaces, in the asreceived state (HV = 2000 MPa) by electrical discharge machining (EDM). Before texturing, the samples were further ground and then mirror-polished using 1 μm diamond paste obtaining a final roughness Rq = 80 nm. The texturing method used to generate the topographic patterns was photochemical etching [15]. Patterns with different geometries were used to texture the surfaces: circular patches, straight lines and chevrons. Figure 2 shows the geometry of the patterns and the sliding orientations for the lines and chevrons; Table 1 presents the dimensions for all the texture patterns used in this study. Note that, for the patterns containing circles, px = py. Reciprocating tests were used to evaluate the behaviour of the textured surfaces in lubricated sliding, giving variable sliding velocity during each stroke of the d
px
w py w
2. Theoretical Background circles If two surfaces are sliding relative to each other with a certain velocity in the presence of a lubricant fluid, a hydrodynamic film may form spontaneously between them and if this is sufficiently thick then hydrodynamic or full-film lubrication prevails. In lubricated sliding, surface textures can influence lubrication mechanisms, leading to changes in friction and wear. For full-film lubrication the empirical understanding of the mechanisms involved in the tribology of textured surfaces is mainly derived from investigations of hydrodynamic seals [12, 13] or from the simulation of hydrodynamic bearings [7, 14].
l
py
lines
chevrons
Sliding orientation (SO)
perpendicular
parallel
perpendicular
parallel
45°
Fig. 2. Microgeometry and sliding orientation of samples
458
A.A.G.Bruzzone and H.L.Costa / Procedia CIRP 12 (2013) 456 – 461 Table 1. Dimensions (in μm) of the features in the texture patterns; NA = not applicable Pattern
Circles
Lines
Chevrons
Sample
SO
d
py
w
C1
NA
70
100
NA
l
C2
NA
60
40
NA
NA
NA
C3
NA
120
460
NA
NA
NA
C4
NA
70
510
NA
NA
NA
C5
NA
45
125
NA
NA
NA
L1
Par.
NA
340
60
NA
NA
NA
NA
L2
45
NA
340
60
NA
NA
L3
Per.
NA
340
60
NA
NA
L4
Per.
NA
200
40
NA
NA
EV0
Par.
NA
NA
40
70
200
EV90
Per.
NA
NA
40
70
200
Table 2. Normal loads and corresponding Hertzian contact pressures and elastic contact widths Load (N)
Contact width (μm)
12.3
82
Contact pressure (MPa) 9.4
22.1
110
12.6
31.9
132
15.1
41.7
150
17.3
51.5
167
19.2
61.3
182
21.0
71.1
197
22.6
80.9
210
24.1
test, which enabled changes in the lubrication regime to be seen in a single test. Sinusoidal linear motion of the samples with simple harmonic motion was produced by a crank mechanism, with a stroke length of 22 mm and a frequency of 0.55 Hz. The contact area was immersed in additive-free mineral oil with a dynamic viscosity of 1.5 Pa s at 20oC. The counterbody was a mirror-polished aluminium cylinder with diameter of 200 mm and length of 16 mm. Sliding direction was perpendicular to the axis of the cylinder, so that the contact was linear, transverse to the direction of motion. Each sample was tested with 8 different normal loads, which resulted in different contact pressures and contact widths (Table 2). The variables continuously measured during each test were the friction force and the electrical capacitance between the two sliding bodies, through the lubricant film. A running-in period was observed from changes in the film capacitance; data were collected only within the subsequent steady-state regime, for a total of 150 sliding cycles for each sample. Points from consecutive cycles corresponding to the same translational velocity were then averaged, to compute mean values of friction force and film capacitance for an average stroke.
From the instantaneous friction force, displacement and velocity of the counterbody [8], the energy dissipated by friction during each stroke, Ef, the corresponding maximum Pmax, and average power Pav, lubricant film thickness TL, and friction force Ff, were computed. Square areas of 2 x 2 mm2 were acquired for each sample every 30 cycles using a laser interferometer, UBM 3D Autofocus, with sampling steps dx = 1 μm and dy = 10 μm along the x and y axes. Two procedures were used to determine the reference surface, or datum: Gaussian filtering with a cut-off c = 100 μm, and least mean square plane fit. Due to the plane form of the samples and the structure of the patterns, the measured differences were negligible, and the second procedure, which maintains the original dimension of the acquired area and is insensitive to the discontinuities due to the pockets, was chosen. The sampled data were treated as gray level images with black pixels (level 0) corresponding to valleys, and white pixels (level 1) to summits. Segmentation was made by identifying the pockets and groove regions using a threshold according to the Otsu's method, which minimizes the intra-class variance of the black and white pixels. The isolated black pixels, corresponding to minima, and the connected black regions with areas less than 80 μm2, were removed successively, and the remaining regions were morphologically dilated with a disk shaped structuring element with a diameter of 10 μm. For each region of the sample a set of parameters was computed (Table 3) and their average considered for the purpose of surface characterization. Figure 3 shows the main concept underlying the computation of the region parameters, particularly the ellipse with normalized second central moments, its principal axes and the minimum depth.
Fig. 3. Computation of surface parameters for an individual region
459
A.A.G.Bruzzone and H.L.Costa / Procedia CIRP 12 (2013) 456 – 461 Table 3. Definition of the surface parameters Units
Definition
A
μm2
Area projected on the reference surface by the region
A0
μm2
Area of the region projected on the reference surface with negative deviation (z < 0)
C zmin
μm
Minimum depths of the region
μm
Equivalent Diameter of the circle with the area A
e1
μm
Major Axis Length of the ellipse with the same normalized second central moments
e2
μm
Minor Axis Length of the ellipse with the same normalized second central moments Eccentricity, the ratio e2/e; 0 corresponds to a circle, 1 to a line
E °
μm
S
V0 zo
Table 4. Mean values of considered surface parameters Texture
Coverage, ratio between the area of the regions and the sample area
deq
P
highest value and provided information on the dimension of the single region. Coverage C could capture the different distribution of circles over the surface when associated to e2.
Orientation angle between the x-axis and the major axis of the second-moments ellipse Perimeter of the region Solidity ratio between the areas of the region and its convex hull, i.e. a measure of the convexity of the single pattern
μm3 μm
Volume under the reference surface of the region Mean depth the region under the reference surface
Due to their definitions, some of the computed surface parameters are correlated; in particular positive correlations between the perimeter (P), the length of the major axis (e1), the areas (A and A0), the volume under the reference surface V0 and the equivalent diameter deq was observed. Also the orientation showed negative correlation with many other parameters of the examined structured surfaces. Therefore the study focused on a subset of parameters that convey important information on texture and tribological phenomena, precisely: V0, , e1, C and S. Noticeably the orientation depends on the relative orientation between the surface texture and the sliding direction. 4. Results and Discussion Table 4 reports the mean values of the surface parameters considered in this analysis. The volume under the reference surface, V0, discriminated between the chevron texture and the other textures. Orientation classified distinctly the lines and the chevrons, while for the circles no apparent trend was seen. The length of the axis e2 was clearly proportional to the circle diameters and the widths of the lines; for chevrons, e2 assumed the
C1 C2 C3 C4 C5 L1 L2 L3 L4 EV0 EV1
V0 (μm3) 20866 11607 5375 10511 2292 411674 411674 411674 96933 209082 209082
(°) -21.77 7.85 27.08 -12.81 -7.79 1.64 -43.36 -88.36 -89.55 -7.37 97.37
e1 (μm) 21 19 32 20 13 587 587 587 519 93 93
C
S
0.172 0.414 0.236 0.014 0.064 0.157 0.157 0.157 0.132 0.193 0.193
0.98 0.98 0.96 0.98 0.98 0.86 0.86 0.86 0.82 0.71 0.71
Table 5. Mean values of the energy absorbed by friction Ef (N mm)
Sur. C1 C2 C3 C4 C5 L1 L2 L3 L4 EV0 EV90
Normal Load 12.3 22.1 17.1 19.6 16.1 18.6 15.9 18.1 16.1 18.4 14.6 17.5 17.5 20.3 16.8 19.9 15.6 18.4 17.1 19.9 21.4 20.4 16.9 19.5
(N) 31.9 21.5 20.2 20.2 20.3 19.7 22.9 23.3 20.8 22.6 22.8 25.6
41.7 23.7 22.2 22.1 22.4 21.6 25.5 26.4 23.0 25.0 25.0 24.0
51.5 25.7 23.8 23.7 24.3 23.5 26.1 30.4 25.1 28.2 27.0 26.0
61.3 27.8 25.5 25.3 26.3 25.4 31.6 35.5 26.6 33.0 29.1 27.9
71.1 29.8 26.8 26.7 28.4
80.9 31.6
37.5
25.5 30.5 28.6 38.4
35.9 31.0 30.1
31.1 40.4 33.5 32.4
Table 6. Mean values of the maximum power Pmax (N mm s-1)
Sur. C1 C2 C3 C4 C5 L1 L2 L3 L4 EV0 EV90
Normal Load (N) 12.3 22.1 31.9 39.1 44.4 50.3 37.2 43.2 46.9 35.6 41.7 47.4 37.7 43.4 48.1 34.3 41.0 45.8 39.7 46.7 52.6 38.3 44.8 52.0 36.4 42.9 48.8 39.1 46.0 52.2 50.1 47.7 53.9 38.6 44.9 59.4
41.7 54.5 51.8 51.4 52.5 50.5 58.3 56.9 53.0 57.3 59.1 55.5
51.5 59.5 55.7 56.0 57.7 54.4 61.1 62.4 57.7 62.1 63.9 59.5
61.3 65.5 60.3 59.2 61.5 60.2 69.2 67.7 60.6 69.6 68.3 63.7
71.1 69.7 62.1 63.1 65.8
80.9 73.1
79.4
58.7 72.9 65.7 80.2
71.0 73.4 67.3
69.2 75.5 78.9 71.6
Table 7. Mean values of lubricant film thickness TL (μm)
Sur. C1 C2 C3 C4 C5 L1 L2 L3 L4 EV0 EV90
Normal Load (N) 12.3 22.1 31.9 16.9 13.8 10.8 14.4 10.5 10.0 14.2 10.4 8.3 15.7 11.3 9.1 20.2 14.8 11.8 25.0 17.5 13.9 25.3 18.3 13.0 19.3 13.9 10.7 25.1 17.6 13.8 11.5 12.8 10.0 16.7 12.1 7.0
41.7 9.0 7.5 7.1 7.4 10.0 11.3 10.7 9.0 11.5 8.5 7.9
51.5 7.9 6.6 6.3 6.4 8.7 9.8 9.5 7.8 9.1 7.4 7.0
61.3 6.8 7.7 5.6 5.9 8.1 8.3 8.1 7.2 7.0 6.6 6.2
71.1 6.1 6.8 5.0 5.1
80.9 5.3
5.3
4.4 4.4 5.8 4.9
7.3 6.1 5.6
5.5 6.4 5.1 4.0
460
A.A.G.Bruzzone and H.L.Costa / Procedia CIRP 12 (2013) 456 – 461
The tribological parameters, namely the energy absorbed by friction Ef, the friction power parameters Pmax and Pav, the lubricant film thickness TL, and the friction force Ff, generally had a positive trend with load for all the textures. The mean values of the tribological parameters Ef, Pmax and TL, for the considered normal loads and surface microgeometries are reported in Tables 5, 6, and 7. More interesting results emerge from the analysis of the relationships between the surface parameters and the tribological parameters. In general for small and medium loads (up to 61.3 N), the correlation between surface and tribological parameters was not very strong. The correlation coefficients between the energy dissipated by friction during each stroke, Ef, and the surface parameters show that a strong dependence of Ef on the considered surface parameters cannot be established; however a moderate correlation between Ef and the solidity S is evident. Analogous considerations hold for the relationships between maximum power Pmax, average power Pav, friction force Ff, and the surface parameters. Particularly, energy and power parameters are positively correlated to the perimeter of the regions P, the major axis length e1 of the second central moment ellipse, and the volume V0 under the reference surface of the regions. Considering lubricant film thickness, TL, a reasonable dependence of the thickness on the major axis length, e1, is observed for nearly all the normal loads (Table 8). Figure 4 shows how the surfaces patterned with circles, lines and chevrons, are classified by e1, and how TL decreases with increasing normal loads. Load 12.3 N
Film Thickness [ m]
22.1 N 31.9 N 41.7 N
51.5 N 61.3 N
e1 Major Axis Length [ m]
Fig. 4. Maximum film thickness as a function of major axis length and normal load; c = circles, l = lines, ev = chevrons
Table 8. Correlation between the maximum lubricant film thickness TL, and the surface parameters for low and medium loads Par.
Normal Load (N) 12.3
22.1
31.9
41.7
51.5
61.3
C
-0.35
-0.37
-0.22
-0.33
-0.31
0.08
e1
0.80
0.79
0.70
0.75
0.73
0.50
-0.53
-0.55
-0.46
-0.52
-0.43
-0.17
P
0.76
0.78
0.65
0.73
0.72
0.47
S
-0.05
-0.24
0.06
-0.20
-0.18
0.09
V0
0.51
0.58
0.38
0.49
0.56
0.45
For higher loads (71 and 81 N) stronger correlations could be observed. Friction force Ff and maximum film thickness TL are negatively correlated to the orientation between the x-axis and the major axis of the angle second-moments ellipse, namely these tribological parameters depends on the sliding direction. Using the correlation results and the mean values of the tribological parameters it is possible to discuss functional performance for each group of textures. 4.1. Circles The hydrodynamic lubrication of patterns containing circles was sensitive to the area coverage C. There was a maximum of film thickness for an intermediate coverage value and then thickness decreased. Two competing effects occur when C increases. On the one hand, there is an increase in the hydrodynamic pressure; on the other, there is a reduction in the contact area, which may result in the maximum observed for all loads. Another possible effect is the distribution of the circles, since a large number of small pockets might have the same coverage C as a smaller number of large pockets. A clear trend could be seen for the effect of V0 on thickness, although the sample with the smallest value (C5) presented the thickest films for all loads, which might be associated with the smaller diameter of the pockets for this sample. Pockets with larger diameters are less likely to be contained within the contact width, particularly in this study, where the contact widths are narrow. This hypothesis illustrates the importance of the distribution of the features besides coverage. Friction was not much affected by the texture, probably because the tests were performed within the hydrodynamic regime. When a full film separates the sliding surfaces, friction occurs due to shearing within the lubricant, and therefore would not be expected to be very sensitive to the texture. The energy dissipated by friction increased with the volume under the reference plane V0.
A.A.G.Bruzzone and H.L.Costa / Procedia CIRP 12 (2013) 456 – 461
4.2. Chevrons The orientation of the chevrons relative to the sliding direction affected the tribological behaviour. When the vertices of the chevrons were oriented in the sliding direction (EV0), the film thickness was higher than when they were oriented perpendicular to the sliding (EV90). The energy dissipated by friction and the friction power were also larger when the vertices of the chevrons were oriented in the sliding direction. For the chevrons, it is believed that the side lands restrict lateral flow and thus increase the hydrodynamic pressure. When the vertex is aligned in the sliding direction, the side restrictions act in conjunction to direct flow towards the region of high hydrodynamic pressure, which further increases pressure. This effect is less significant when the vertices are aligned perpendicularly to the sliding direction. 4.3. Lines The sample with lines oriented perpendicularly to the sliding direction (L1) showed worse behaviour than the other two orientations. However, for higher loads, the difference between the samples tended to decrease. For the highest normal loads, film thickness for the perpendicular lines (L3) was slightly higher. The difference between the samples L3 and L4 might be due to other variables, such as coverage C. The variation of friction energy and power with orientation was consistent with the thickness results. This difference in behaviour for different loads could be explained in terms of the variation of contact width. For relatively high loads and, consequently, large contact widths, all orientations give similar proportions of textured areas inside the contact, resulting in similar films for all orientations. For narrow contact widths, perpendicular lines may result in a smaller proportion of texture inside the contact, which may explain why the film was thinner than for the other two orientations. For the highest load, the thicker film measured for L1 and L3, perpendicular to the sliding direction, might be due to a tendency of the lines to channel the lubricant away from the contact as they become oriented towards the sliding direction, although the effect was very slight. Costa and Hutchings found that this channelling effect can be very significant for situations where the contact width is much larger than the size of the features, such as in lubricated strip drawing using textured dies [8]. 5. Conclusions The study here reported outlines the possibility of developing parameters to capture the tribological
461
function of structured surfaces under conditions of fluid film lubrication. It provides a methodology to select the appropriate surface functional parameters. An appropriate use of surface analysis tools such as 3D morphological filters enabled the experimental assessment of the relationships between tribological parameters and the microgeometry of textured surfaces. This research outlines how the textured surfaces behave in two different modes that depend on the normal load. The results emphasize the importance of the orientation of the texture in respect to the sliding direction, and of the volume contained in the pockets/grooves. References [1] Bruzzone, A.A.G., Costa, H.L., Lonardo, P.M., Lucca, D.A.., 2008. Advances in engineered surfaces for functional performance, CIRP Annals - Manufacturing Technology 2008; 57/2, p.750 . [2] Hutchings, I.M., 1992. "Tribology - friction and wear of engineering materials". London: Arnold. [3] Williams, J.A., 1994. Engineering tribology, Oxford: Oxford University Press. [4] Kim, D.E., Cha, K.H., Sung, I.H.., 2002. Design of surface micro-structures for friction control in micro-systems applications, Annals of the CIRP 2002; 51/1, p.495. [5] Zumgahr, K.H., Mathieu, M., Brylka, B., 2002. Friction control by surface engineering of ceramic sliding pairs in water, Wear 2002; 263, p. 920. [6] Pettersson, U., Jacobson, S., 2003. Influence of surface texture on boundary lubricated sliding contacts, Tribology International 2003; 36, p.857. [7] Wang, J.D., Chen, H.S., Han, Z.L., Chen, D.R., 2010. Investigation of milli-scale dimples on planar contact lubrication, Tribology Transactions 2010; 53, p.564. [8] Costa, H.L., Hutchings, I.M., 2008. Hydrodynamic lubrication of textured steel surfaces under reciprocating sliding conditions, Tribology International 2007; 40, p.1227. [9] Stout, K.J., Sullivan, P.J., Dong, W.P., Mainsah, E., Luo, N., Mathia, T., Zahouani, H.., 1993. The development of methods for the characterization of roughness in three dimensions, Report EUR 15178 EN 1993. [10] Jiang, X., Scott, P.J, Whitehouse, DJ,Blunt, L., 2007. Paradigm shifts in surface metrology. Part II. The current shift, Proc. R. Soc. A. 2007; 463, p.2071. [11] De Chiffre, L., Lonardo, P., Trumpold, H., Lucca, D.A., Goch, G., Brown, C.A., Raja, J., Hansen, H.N., 2000. Quantitative characterisation of surface texture, Annals of the CIRP 2000; 49/2, p.635. [12] Hamilton, D.B., Walowit, J.A., Allen, C.M., 1966. A theory of lubrication by micro-irregularities, J Basic Eng-Trans ASME 1966, March, p.177. [13] Etsion, I., Burstein, L., 1996. A model for mechanical seals with regular microsurface structure, Tribology Transactions 1996; 39, p.677. [14] Brizmer, V., Kligerman, Y., Etsion, I., 2003. A laser surface textured parallel thrust bearing, Tribology Transactions 2003; 46, p.397. [15] Madou, M.J., 2002, Fundamentals of microfabrication.2nd ed. Florida: CRC Press.
