- Email: [email protected]
copper-graphite sliding electrical contact materials
Author’s Accepted Manuscript Wear phenomena and tribofilm formation of copper/copper-graphite sliding electrical contact materials M. Grandin, U. Wiklund www.elsevier.com/locate/wear
PII: DOI: Reference:
S0043-1648(17)31477-1 https://doi.org/10.1016/j.wear.2017.12.012 WEA102319
To appear in: Wear Received date: 16 October 2017 Revised date: 15 December 2017 Accepted date: 15 December 2017 Cite this article as: M. Grandin and U. Wiklund, Wear phenomena and tribofilm formation of copper/copper-graphite sliding electrical contact materials, Wear, https://doi.org/10.1016/j.wear.2017.12.012 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Wear phenomena and tribofilm formation of copper/copper-graphite sliding electrical contact materials M. Grandin*, U. Wiklund Tribomaterials group, Applied Materials Science, Uppsala University Ångström Laboratory, Box 534, SE-751 21 Uppsala, Sweden *Corresponding author: E-mail address: [email protected]; Tel.: +46 18 471 3069
Abstract Copper-graphite composites in reciprocating sliding against copper are investigated with regards to friction, wear and contact resistance. The tribological and electrical evaluation is complemented with surface analysis. It is shown that the presence of graphite in the composite greatly reduces the coefficient of friction and wear. However, the amount of graphite is not critical and only has a minor influence on friction and wear. It is observed that the coefficient of friction is slightly lower, while the wear rate is slightly higher, in pure mechanical tests than in tests with current. The contact resistance is greatly reduced by an increase in the copper content in the composite. Chemical analysis of the tribofilm that forms on the copper surface shows that it consists of graphite as well as Cu2O. It is shown that the mating couple with the highest amount of oxide in the tribofilm also has the lowest contact resistance. Hence, it is concluded that oxides are not necessarily detrimental for the contact resistance as long as there is unoxidized copper available. Novel cross-section techniques and images of the tribofilm contribute to a deeper understanding of how sliding electrical contact surfaces are affected by current and sliding motion. Keywords: Tribofilm, electrical contact, copper-graphite, contact resistance, friction
1. Introduction Continuously sliding electrical contacts are found, for example, where current or signals are transferred between one stationary and one rotating component. As such, they are tribological systems where friction and wear influence the performance. Carbon and graphite brushes are used for power transfer in everything from large generators to household appliances. Adding a metal, such as copper, will lower the resistivity and increase the current carrying capabilities [1]. Such metal-graphite composite materials are used to transfer power and signals in e.g. slip-ring assemblies and in locomotive pantograph strips for current collection. A satisfactory transfer of current, with low losses and low noise levels, are generally prioritized. Still, there is a lot to gain from improving also the tribological properties. In this paper, we present results on copper-graphite composite materials sliding against copper. Friction, wear and contact resistance are measured and special attention is paid to the contacting surfaces after testing, looking for material transfer, surface modifications and formation of tribofilms. Although the relationship between contact resistance and copper content in composite is rather straightforward, the correlation to friction and wear is more complicated. Both friction and wear benefit from small real contact areas, which can for example be achieved with hard materials. Although increasing the hardness usually has an impact on the friction and wear, it is important to remember that these parameters are not intrinsic to a specific material. For example, it is not necessarily the hardness of the materials that you start out with that makes a difference, but rather the tribofilm that is formed on the surfaces during sliding. It will be influenced by the specific circumstances during sliding, surrounding atmosphere, normal load etc. and its
nature is decisive for the friction, wear [2] and contact resistance. An optimal low friction situation is one where a thin tribofilm with low shear strength is formed and maintained on top of a hard surface. In this way, there will be benefits from both a small real contact area from the hardness and a low shear strength from the soft and easily sheared tribofilm. The formation of such tribofilms can be promoted by adding solid lubricants such as graphite, MoS2 or WS2. Due to their layered structure, these materials have low shear strength. In the case of graphite, however, the layered structure is necessary, but not sufficient. To achieve low friction, adsorption films are also necessary [3]. As system parameters, friction, wear and contact resistance need to be studied under specific circumstances. However, in the literature, focus is most often on either the electrical or the tribological properties. Transferring a current over a tribological interface may alter the friction and wear behaviour through increased temperature as the current is restricted to pass through the conducting spots and through a preferential direction of material transfer [4]. Still, copper-graphite materials are often studied without a transfer of current although electrical contacts are mentioned as a likely application [5–10]. Other studies apply current during testing, but do not measure the voltage drop [11–16] which hinders linkage between tribological and electrical performance to be studied. Some early work done on copper-graphite that is worth mentioning is that of J. Casstevens and co-workers [17–19] and Johnson and Moberly [20]. A lot of relevant knowledge was summarized by Rohatgi in 1992 [1] although his review includes other metal-graphites than coppergraphites as well as other applications than electrical contacts. More recent work focus on improvements related to different forms of graphite [5–8,21] or related to the manufacturing process [11,22,23], all of which have been evaluated against rotating steel. Correlations of the tribological properties and the amount of graphite in the composite show that increasing the vol.% of graphite up to 20% will lower the friction and wear, however, the positive effects on the tribological properties will not continue to increase with a further increase in the amount of graphite [1]. Scratch-testing of copper-graphite composites with varying amount of graphite up to 30 vol.% against a spherical diamond tip shows that the wear mechanism change from ploughing to microcutting with increasing load [24]. The same study also conclude that 6-20 vol.% of graphite is optimum for low shear forces, while more graphite result in particles coming loose, resulting in higher levels of wear. A few papers considering both tribological and electrical performance can be found in the literature, e.g. Bucca and Collina who make thorough investigations of the pantographcatenary system [25,26]. In addition, some studies of copper-graphite against copper or copperalloys can be mentioned; Cho et al. varied the amount of copper in the composite and the mechanical load and found that the voltage drop was minimized at 60 vol.%. and that the friction decrease with a decrease in copper content and mechanical load [27]. Shin and Lee found a correlation between brush wear and brush temperature [28]. The present work tries to increase the knowledge of the mechanisms behind such results as neither Bucca and Collina nor Cho, nor Shin and Lee investigate the contacting surfaces in detail.
2. Experimental 2.1. Test set-up and procedure Triboelectrical evaluation was performed in reciprocating sliding with two horizontal, crossed cylinders, at right angles, see Figure 1a. The voltage drop over the contact interface was measured using the four-point probes method where current-carrying electrodes are separated from voltagesensing electrodes avoiding voltage drop contributions in wires and other contact points than the one targeted.
Figure 1. a) Illustration of the crossed-cylinders and four-point probes configuration for voltage drop measurements. b) Test setup apparatus.
The cylinders are 20 mm long and 10 mm in diameter and the upper stationary cylinder was pressed against the lower moving cylinder using a spring applying 10 N normal load, see Figure 1b. A current controlled power supply was used to feed a direct current of 5 A through the contact, with the upper cylinder configured as cathode and the lower as anode. A stroke length of 5 mm and a 3 Hz frequency was used for 30 000 strokes (expect for the copper-copper reference test that was interrupted after 5000 strokes). Both the normal load and the friction force were measured during testing, as was the voltage drop from which the contact resistance was calculated assuming a constant current. Testing was performed without lubrication and at ambient lab conditions, the latter with variations in temperature and relative humidity small enough not to have any significant influence on the results.
2.2. Materials Copper cylinders were used in the stationary, cathode position. The copper cylinders are made from oxygen free copper with a purity of 99.95%. They were circumferentially polished prior to testing to an average Ra value of about 50 nm. The hardness of this material is 1.9 GPa and the elastic modulus is 140 GPa. As the moving, anode specimen, three different commercially available copper-graphite composite materials were used. These materials are pressed from powders and heat-treated. Some properties of these materials are listed in Table 1, as given by the manufacturer1. Note that the composite materials are named after the vol.% metal. The hardness is measured according to DIN IEC 413.303 (a method particularly developed for brushes, performed as a Rockwell test using a 10 mm ball and 60 kg load) and the resistivity is measured according to DIN IEC 413.402. After pressing, the composites had achieved a density that was 85% (Cu6), 95% (Cu26), and 75% (Cu70) of the maximum. The SEM images in Figure 2 give an indication of the distribution of metal and graphite in the composites. The composite nature of these specimen makes polishing challenging but all cylinders were circumferentially polished to a Ra value of approximately 200 nm. Nanoindentation hardness and elastic moduli were also determined separately for the two phases within the composites. On average, the graphite has HIT=0.6 GPa and EIT= 15 GPa and the copper HIT= 1.9 GPa and EIT=110 GPa. Although the hardness of copper is higher than the hardness of graphite, the softest of the three composites, according to the DIN IEC 413.303 method, is the one with the largest amount of copper. Table 1. Properties of the copper-graphite composites. Vol.% metal Hardness HR10/60 Resistivity (µΩm) Density (g/cm3)
Cu6 6 75 12.0 2.3
Cu26 26 85 0.82 3.2
Cu70 70 50 0.05 5.8
Figure 2. Polished copper-graphite composite surfaces illustrating the distribution of copper (bright) and graphite (dark).
2.3. Analysis The surfaces were analysed after testing with the objective of correlating the surface appearance with the tribological and electrical performance. White light interference profilometry for investigation of the worn composite surfaces was performed on an optical profiling system (WYKO NT1100, Veeco, USA). The length of the wear scar and the wear volume was also measured and used for calculating a specific wear rate. Surfaces were imaged in an SEM (Merlin, Zeiss,
1
Carbex AB, Sweden
Germany) and analysed with EDS (Aztec software). This SEM instrument is equipped with a FEG electron gun and X-Max Silicon drift detector. EDS compositions of the graphite wear scar (acquired using 5 kV acceleration voltage) and tribofilm on the copper cylinder (10 kV) are presented in at.%. The EDS composition outside the wear scar is also included to show the changes in composition. These compositions are of course numerically different from the vol.% compositions used during manufacturing and for naming the composites (c.p. Table 1). As EDS is not optimal for composition analysis of light elements, these results are mainly intended for qualitative comparisons. A broad ion beam (BIB) argon milling system (Ilion II, Gatan, USA) was used for milling cross-sections in the tribofilms formed on the copper cylinders. AGAR silver glue (thin silver flakes in methyl isobutyl ketone suspension) was used to mount the sample on a titanium blade for the milling operation. Raman spectroscopy was performed (inVia micro-Raman system, Renishaw, UK) using a 532 nm laser at a power below one mW. A nanoindenter (UNHT, CSM, Switzerland) was used for the hardness results presented above. A Berkovich tip was used and the Oliver & Pharr method [29]. For indentations in the copper phase in the copper-graphite the average load was 1450 µN and the depth 165 nm. For the graphite, the average load was 500 µN and the depth 190 nm. The results are in each case an average of at least 20 indentations
3. Results and discussion 3.1. Contact resistance Adding more metal to the copper-graphite composite results in a lower contact resistance, see Figure 3. This is an expected result as copper has a lower resistivity than graphite. Five identical tests were conducted and the standard deviation is shown as a shaded area around the mean curve. The contact resistance is increasing during the test period of 30 000 strokes, for Cu6 and most prominently for Cu70. From this observation it is clear that the contact situation changes over time; either the true contact area has been reduced or the materials in the contacting surfaces have been modified. The contact resistance for the three composite materials, approximated to a mean value over the 30 000 strokes, is plotted against the copper content in the inset in Figure 3. It turns out that, within this copper content interval, this relationship is roughly linear. It can be observed that the initial contact resistance behaviour differs greatly between Cu6 and Cu26. While Cu6 starts at a value much higher than its contact resistance after stabilization, Cu26 starts at a much lower value. In Cu6, which has a larger amount of graphite, some sliding is necessary to spread metal over a larger part of the contact surface and provide a low contact resistance. Initially, there are isolated spots of copper, which is not ideal for the transfer of current. For Cu26 there is enough copper to provide a low contact resistance already at the start, and the gradual increase is believed to be due to accumulation of insulating oxides at the interface.
Figure 3. The contact resistance increases gradually with sliding distance for Cu6 and Cu70, but it is almost constant for Cu26. The right inset shows how the contact resistance, averaged over the 30 000 strokes, decreases linearly with increasing vol.% of metal in the composite.
3.2. Coefficient of friction The coefficient of friction on the other hand does not relate to the amount of metal in the composite in a simple linear manner; see Figure 4 and Figure 5 As for the contact resistance, the standard deviation is shown as a shaded area around the mean curve. The friction is stable from around 15 000 strokes and hence a mean value of the last 15 000 strokes has been used to illustrate the mean steady state friction and its standard deviation in Figure 6. During the first 5000 strokes, a larger standard deviation, and hence large spread between tests, is observed, before they stabilize at very similar friction values. The inset in Figure 6 shows the friction of a copper-copper reference test, which makes it obvious that the presence of graphite is of great importance for the friction properties of this system. Among the composites tested in this study Cu26 results in the lowest coefficient of friction. Cu26 is the hardest of three composites, which could influence the friction provided that the shear strength in the interface is the same in all three cases. The coefficient of friction for the composite with the least amount of graphite has a mean value between the other two, but it can also be seen that the variation among identical tests is much greater for this composite. The reason for this is believed to be that the graphite is not evenly distributed in the copper matrix, which means that the amount of available graphite can differ from one test to another. For all three composites, the coefficient of friction is lower in tests without current than with current, although the high standard deviation for Cu70 makes these results uncertain. It can be estimated that the current induced heat in the contact spot dominates over friction induced heat. For Cu6 and Cu26 the voltage drop is above the softening voltage for copper (120 mV [30]), so it is likely that the additional heat due to current increases the size of the of a-spots. This will in turn increase also the real contact area and result in an increased friction.
Figure 4. The coefficient of friction is lower for Cu26 than for Cu6. For both composites, the friction is higher with than without current.
Figure 5. For the Cu70 composite, the variation between tests is larger than for the other two composites. A Cu-Cu reference test is shown in the inset.
Figure 6. Mean values of the steady state friction and its standard deviation for all three composites shows that Cu26 gives the lowest friction and Cu70 the highest standard deviation. For all composites, testing with current results in a slightly higher friction coefficient.
3.3. Effects on the composite cylinders In this section, results concerning the composite cylinder will be presented, first surface analysis of the wear scar and then the wear rate. The SEM images in Figure 7 give an idea of the appearance of the wear scar of the different composites. The difference in metal content in the composite is reflected also in the wear scar and hence the surfaces of the wear scars look very different from each other.
Figure 7. The wear scars in the different composites at two different magnifications (5 A).
There is no straightforward relationship between specific wear rate of the composite and its composition as can be seen in Figure 8. The variation within repeated tests is fairly high as indicated by the error bars; this is especially true for the Cu70 composite. Focussing on comparing Cu6 and Cu26, we see two trends; feeding 5 A current through the contact causes less wear of the composite than if no current is fed and more metal in the composite results in lower wear rate of the composite. The overall composition in the wear scar on the composite cylinders, tested with and without current, are shown in Figure 9, together with that of the untested composite for comparison. After testing, the Cu to C ratio has increased, i.e. there is an enrichment of metal as a result of sliding. This indicates that there is a preferential wear of graphite and/or that smearing of the malleable copper will cover part of the graphite. This is in accordance with the brighter shade inside than outside the wear scars in Figure 7. There is also an enrichment in oxygen after testing. Sliding will remove the oxidized surface layer, which uncover pure copper. This copper will, in turn, oxidize and with time, more and more oxide will accumulate in the wear track. In most cases, there appears to be more metallic copper in tests with than without current. The small additional heat in the contact, as a current is passing, may soften the copper slightly more and cause it to deform and cover the surface a little bit more, resulting in an enrichment of copper. Weakening of the graphite phase due to increased heat with current is one explanation of increased composite wear with current described in the literature [31]. Another possible mechanism could be that softening of the copper due to heat from the current, see section 3.2, will weaken the composite and result in wear. In our case, however, more heat is not associated with more wear. Instead, perhaps the observed increased amount of copper in the surface could counteract this weakening of the composite, resulting in a lower wear rate with than without current. For Cu6, the Cu to O ratio is below two, which indicates that all the copper in the surface is oxidized (assuming Cu2O, see the Raman results below).
Figure 8. The specific wear rates of the composite cylinders. The scatter between tests is large, particularly for Cu70.
Figure 9. Surface composition in the wear scar of the composite cylinders. Compositions of the composites, measured in the same way, are shown as reference. Sliding increases the relative amount of copper and oxygen.
3.4. Effects on the copper cylinders Tribofilms form on all copper cylinder surfaces, see Figure 10. For convenience, all images etc. are denoted by the counter copper-graphite cylinder composition in parentheses; (Cu6), (Cu26) and (Cu70), although it is the copper cylinder that is analysed. All tribofilms show a similar appearance although a slightly darker shade is seen for (Cu6) indicating a higher abundance of light elements like carbon and oxygen.
Figure 10. The tribofilms on the copper cylinder surface. The contrast is slightly different indicating a difference in composition. These examples are from tests at 5 A current but no differences are seen when comparing to tests without current.
From the EDS mapping shown in Figure 11 it is clear that for (Cu26) and particularly for (Cu6) there is an accumulation of carbon in the tribofilm which is not seen for (Cu70) where the amount of carbon on the surface is barely more than the natural carbon contamination always present on surfaces. The maps of copper also differ significantly. For (Cu6), and somewhat for (Cu26), the carbon and oxygen in the tribofilm is thick enough to shield the underlying copper. Mapping the oxygen concentration shows that the oxygen is more evenly distributed for (Cu70) than for the other two. The EDS composition of the tribofilm, as measured in the central part of each mark, is presented in Figure 12. As for the EDS analysis of the composite wear scar, for (Cu6) the Cu to O
ratio is below two, indicating that most of the copper is oxidized, while for (Cu70) metallic copper is present as well. Comparing the tribofilm composition with the wear scar composition, it is evident that the compositions are very similar on the two sides of the contact interface and there is arguably a tendency for higher Cu content with than without current.
Figure 11. EDS mapping of the tribofilms on the copper cylinder surface after testing at 5 A current shows that the composition seems to be related to the composition of the copper-graphite composite. It is clear that there is an enrichment of carbon and oxygen in the tribofilm.
Figure 12. Composition of the central region of the tribofilms. More graphite in the copper-graphite cylinder results in more graphite in the tribofilm.
As both graphite and copper-oxides are Raman active, this spectroscopy method can be used to detect the presence of graphite on the surface and to determine if oxides are present, and if so,
which kind they are. At lower wavenumbers five separate peaks can be identified in Figure 13: 114, 124, 148, 216, 528 and 627, and at least the last four of these are associated with Cu2O according to literature [32–36]. No other copper-oxides have peaks in the interval 110-120 cm-1 and the compound(s) giving rise to these peaks is unknown. The peaks above 1000 cm-1 in the Raman spectra can all be associated with disordered graphite [37]. At the very bottom of Figure 13, a Raman spectrum of the copper cylinder is shown and Cu2O is detected. Above that, spectra from the copper and graphite phases of the composite are also included as references where graphite is detected on both phases, and Cu2O only on the copper phase. Above that are spectra from three different tribofilms, produced against the different composites tested at 5 A current. In all cases, there is evidence for Cu2O and graphite. One spectrum from a test without current is also included to further prove the presence of oxide also in that case and to show that there is no significant difference between a test with and one without current. There are however, some differences depending on composition of the mating composite cylinder and this was investigated further by mapping with Raman. Figure 14 shows that the graphite layer covers more of the tribofilm surface with increasing amount of graphite in the mating composite cylinder. It also shows that the highest amount of surface oxide is in the (Cu70) tribofilm. Although the EDS analysis shows the most oxygen for (Cu70) and Raman mapping shows Cu2O in that tribofilm, the Cu70 gives the lowest contact resistance and it is concluded that presence of oxides is not detrimental for the electrical performance, as long as there is also metallic copper present that can transfer the current.
Figure 13. Raman spectroscopy of the tribofilms shows that Cu2O and graphite is present regardless of composite composition and weather being tested with or without current.
Figure 14. Raman spectroscopy mapping showing the distribution of graphite and Cu2O in the tribofilm.
The tribofilm was further analysed by cross-sectioning. Figure 15a illustrates how the cylinder is cut to produce the cross-section. Two reference surfaces are shown in Figure 15b and c. The first one is a cross-section of a copper cylinder that has not been subject to triboelectrical testing, the surface is smooth, and the copper grains are large. The second shows a cross-section of a wear mark from a Cu-Cu reference test that was run for 5000 strokes at 5 A current. Already after such a short running period, there is severe wear, an uneven surface and grain refinement to a depth of several micrometres. Figure 16 shows three examples of tribofilms formed against different coppergraphites in tests without current. Tribofilms formed at 5 A current are not shown but have a very similar appearance, no significant difference can be observed. For all three cases, we see an uneven surface and an uneven tribofilm, sometimes more than 5 µm thick. There is extensive deformation and mechanical mixing of materials with a tendency for vortex-like formations. There is also a grain refinement in the copper below the tribofilm. The deformation and strain, as a result of tribological action, will lower the recrystallization temperature of the copper surface [38,39]. This is the reason for recrystallization, and grain refinement, occuring already at ambient temperature. As this is observed also in tests without current, the heat from friction is apparently enough to give recrystallization and additional heat from the current is not necessary. The tribofilm has an overall darker contrast (more graphite) when there is more graphite in the mating composite, which is consistent with EDS results. As the graphite does not conduct current as well as copper, this difference can be expected to influence the contact resistance so that more graphite in the composite gives a higher contact resistance, which is consistent with the results shown in Figure 3. A closer look at the different components of a tribofilm is presented in Figure 17 where EDS has been used to analyse different spots in a cross-section of a (Cu70) tribofilm formed with 5 A current. The bottom part of the image, below the tribofilm, consists of pure unaffected Cu, although a tiny C peak resulting from surface contamination can be distinguished in the corresponding EDS spectrum. The second spectrum from the top is from a bright contrast area in the tribofilm, which consists mainly of copper, but tiny carbon and oxygen peaks are visible. The next spectrum below that is from a darker contrast area of the tribofilm, having larger amount of carbon and oxygen. A copper surface in air always has a thin natural oxide and this natural oxide is a part of all post-test analysis performed in this study. However, the fact that oxygen is observed throughout the depth of the tribofilm, makes it clear that the oxides observed in post-test analysis is not solely a thin natural oxide. Sometimes there are even darker contrast areas/particles, often found between tribofilm and copper. Apart from Cu and C, also Si and Al peaks are most often detected in these, together with a distinct oxygen peak. As this kind of dark features are neither present in the copper surface after polishing, c.p Figure 15b, nor in self-mated copper wear tracks, c.p Figure 15c, they must originate from the composites. Small amounts of Si and Al are present in the copper-graphite composites, most likely in the form of oxides. Being much harder than the copper, such oxides may wear the copper by abrasion before they become embedded. However, there are no signs of abrasive wear.
Figure 15. a) Illustration of how the cylinder is cut to produce a cross-section transverse to the sliding direction. b) A cross-section of a copper cylinder that has not been subjected to triboelectrical testing. b) A cross-section of a copper cylinder that has been tested for 5000 cycles against another copper cylinder.
Figure 16. Tribofilm cross-sections transverse to the sliding direction. The mating cylinders have different compositions, which is reflected in the tribofilms, but all have similar appearances with very uneven interfaces (10N load, without current).
Figure 17. EDS analysis of a cross-section transvers to the sliding direction showing that there is Si and Al present together with oxygen. The darker the contrast in the tribofilm, the more C and O is detected.
4. Conclusions Copper-graphite materials for sliding electrical contact applications have been evaluated in reciprocating sliding against copper. Testing has been performed both with and without current and the contact resistance is evaluated using the four-point probes method. The wear rate was measured and the surfaces analysed.
More graphite in the composite cylinder results in more graphite on the tribofilm forming on the copper cylinder. Both the composite cylinder surface and the tribofilm are enriched in copper during sliding and even more so if a current is fed through the contact. The tribofilm consists mainly of copper, graphite and Cu2O.
The contact resistance decreases linearly with increasing vol.% of metal in the composite.
The presence of oxides in the tribofilm is not detrimental for the contact resistance, which is shown for (Cu70) where the tribofilm contain the largest amount of oxygen, but the contact resistance is by far the lowest.
The coefficient of friction is similar for all three composites despite the large differences in graphite content. A total absence of graphite results in high friction and wear.
For all three composites, the coefficient of friction is higher, and the wear rate is lower, when a current is fed through the contact.
Increasing the amount of metal in the graphite composite decreases the wear rate.
Acknowledgements The authors would like to acknowledge Carbex AB for supplying materials.
References [1]
P.K. Rohatgi, S. Ray, Y. Liu, Tribological properties of metal-matrix-graphite particle composites, Int. Mater. Rev. 37 (1992) 129–149.
[2]
S. Jacobson, S. Hogmark, Tribofilms – On the crucial importance of tribologically induced surface modifications, in: G.K. Nikas (Ed.), Recent Dev. Wear Prev. Frict. Lubr., Research Signpost, 2010: pp. 197–225.
[3]
R.H. Savage, Graphite lubrication, J. Appl. Phys. 19 (1948) 1–10. doi:10.1063/1.1697867.
[4]
G.F. Dorsey, E.W. Glossbrenner, Sliding contacts for instrumentation and control, in: P.G. Slade (Ed.), Electr. Contacts, Princ. Appl., 2nd ed., CRC Press, Taylor & Francis group, 2014: pp. 1093–1149.
[5]
J. Wang, R. Zhang, J. Xu, C. Wu, P. Chen, Effect of the content of ball-milled expanded graphite on the bending and tribological properties of copper-graphite composites, Mater. Des. 47 (2013) 667–671. doi:10.1016/j.matdes.2013.01.008.
[6]
J. Li, L. Zhang, J. Xiao, K. Zhou, Sliding wear behavior of copper-based composites reinforced with graphene nanosheets and graphite, Trans. Nonferrous Met. Soc. China. 25 (2015) 3354–3362. doi:10.1016/S1003-6326(15)63970-X.
[7]
J. Kováčik, Š. Emmer, J. Bielek, L. Keleši, Effect of composition on friction coefficient of Cu-graphite composites, Wear. 265 (2008) 417–421. doi:10.1016/j.wear.2007.11.012.
[8]
S.F. Moustafa, S.A. El-Badry, A.M. Sanad, B. Kieback, Friction and wear of coppergraphite composites made with Cu-coated and uncoated graphite powders, Wear. 253 (2002) 699–710. doi:10.1016/S0043-1648(02)00038-8.
[9]
K. Rajkumar, S. Aravindan, Tribological performance of microwave-heat-treated coppergraphite composites, Tribol. Lett. 37 (2010) 131–139. doi:10.1007/s11249-009-9499-2.
[10]
W. Ma, J. Lu, Effect of sliding speed on surface modification and tribological behavior of copper-graphite composite, Tribol. Lett. 41 (2011) 363–370. doi:10.1007/s11249-010-9718-x.
[11]
M. Klement, O. Lott, A. Nagel, Synthesis and tribo-electric characterization of coppergraphite- composites with interpenetrating microstructure for sliding contacts, in: 27th Int. Conf. Electr. Contacts, Dresden, 2014: pp. 226–230.
[12]
H. Zhao, G.C. Barber, J. Liu, Friction and wear in high speed sliding with and without electrical current, Wear. 249 (2001) 409–414. doi:10.1016/S0043-1648(01)00545-2.
[13]
X.C. Ma, G.Q. He, D.H. He, C.S. Chen, Z.F. Hu, Sliding wear behavior of copper-graphite composite material for use in maglev transportation system, Wear. 265 (2008) 1087–1092. doi:10.1016/j.wear.2008.02.015.
[14]
G.X. Chen, F.X. Li, L. Dong, M.H. Zhu, Z.R. Zhou, Friction and wear behaviour of stainless steel rubbing against copper-impregnated metallized carbon, Tribol. Int. 42 (2009) 934–939. doi:10.1016/j.triboint.2008.12.011.
[15]
S. Du, F. Zhao, Y. Zhang, Friction and wear behavior of copper-graphite composite material in high-speed sliding with current, Adv. Mater. Res. 487 (2012) 411–415. doi:10.4028/www.scientific.net/AMR.487.411.
[16]
Y.A. Wang, J.X. Li, Y. Yan, L.J. Qiao, Effect of electrical current on tribological behavior of copper-impregnated metallized carbon against a Cu–Cr–Zr alloy, Tribol. Int. 50 (2012) 26–34. http://dx.doi.org/10.1016/j.triboint.2011.12.022.
[17]
J.M. Casstevens, H.G. Rylander, Z. Eliezer, Friction and Wear Characteristics, Wear. 49 (1978) 169–178.
[18]
J.M. Casstevens, H.G. Rylander, Z. Eliezer, Friction and wear properties of two types of copper-graphite brushes under severe sliding conditions, Wear. 50 (1978) 371–381.
[19]
J.M. Casstevens, H.G. Rylander, Z. Eliezer, Influence of high velocities and high current densities on the friction and wear behavior of copper-graphite brushes, Wear. 48 (1978) 121–130.
[20]
J. Johnson, L. Moberly, High-Current Brushes, Part I: Effect of Brush and Ring Materials, IEEE Trans. Components, Hybrids, Manuf. Technol. 1 (1978) 36–40.
[21]
K. Rajkumar, S. Aravindan, Tribological behavior of microwave processed coppernanographite composites, Tribol. Int. 57 (2013) 282–296. doi:10.1016/j.triboint.2012.06.023.
[22]
H. Sarmadi, A.H. Kokabi, S.M. Seyed Reihani, Friction and wear performance of coppergraphite surface composites fabricated by friction stir processing (FSP), Wear. 304 (2013) 1–12. doi:10.1016/j.wear.2013.04.023.
[23]
A. Guillet, E. Yama Nzoma, P. Pareige, A new processing technique for copper-graphite multifilamentary nanocomposite wire: Microstructures and electrical properties, J. Mater. Process. Technol. 182 (2007) 50–57. doi:10.1016/j.jmatprotec.2006.06.019.
[24]
J.K. Xiao, L. Zhang, K.C. Zhou, X.P. Wang, Microscratch behavior of copper-graphite composites, Tribol. Int. 57 (2013) 38–45. doi:10.1016/j.triboint.2012.07.004.
[25]
G. Bucca, A. Collina, A procedure for the wear prediction of collector strip and contact wire in pantograph-catenary system, Wear. 266 (2009) 46–59. doi:10.1016/j.wear.2008.05.006.
[26]
G. Bucca, A. Collina, Electromechanical interaction between carbon-based pantograph strip and copper contact wire: A heuristic wear model, Tribol. Int. 92 (2015) 47–56. doi:10.1016/j.triboint.2015.05.019.
[27]
K.H. Cho, U.S. Hong, K.S. Lee, H. Jang, Tribological properties and electrical signal transmission of copper-graphite composites, Tribol. Lett. 27 (2007) 301–306. doi:10.1007/s11249-007-9234-9.
[28]
W.-G. Shin, S.-H. Lee, An analysis of the main factors on the wear of brushes for automotive small brush-type DC motor, J. Mech. Sci. Technol. 24 (2010) 37–41. doi:10.1007/s12206-009-1135-4.
[29]
W.C. Oliver, G.M. Pharr, An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments, J. Mater. Res. 7 (1992) 1564–1583.
[30]
R. Holm, Electric Contacts; Theory and applications, 4th ed., Springer, Berlin, 1967.
[31]
Y.A. Wang, J.X. Li, Y. Yan, L.J. Qiao, Effect of pv factor on sliding friction and wear of copper-impregnated metallized carbon, Wear. 289 (2012) 119–123. doi:10.1016/j.wear.2012.04.006.
[32]
N. Wang, X. Cao, X. Cai, Y. Xu, L. Guo, Porous cuprite films: facile solution deposition and their application for nitrite sensing., Analyst. 135 (2010) 2106–10. doi:10.1039/c000086h.
[33]
Y. Wang, J. Ghanbaja, F. Soldera, S. Migot, P. Boulet, D. Horwat, et al., Tuning the structure and preferred orientation in reactively sputtered copper oxide thin films, Appl. Surf. Sci. 335 (2015) 85–91. doi:10.1016/j.apsusc.2015.02.028.
[34]
Y.S. Gong, C. Lee, C.K. Yang, Atomic force microscopy and Raman spectroscopy studies on the oxidation of Cu thin films, J. Appl. Phys. 77 (1995) 5422–5425. doi:10.1063/1.359234.
[35]
B. Balamurugan, B.R. Mehta, D.K. Avasthi, F. Singh, A.K. Arora, M. Rajalakshmi, et al., Modifying the nanocrystalline characteristics - Structure, size, and surface states of copper oxide thin films by high-energy heavy-ion irradiation, J. Appl. Phys. 92 (2002) 3304–3310. doi:10.1063/1.1499752.
[36]
Z.H. Gan, G.Q. Yu, B.K. Tay, C.M. Tan, Z.W. Zhao, Y.Q. Fu, Preparation and characterization of copper oxide thin films deposited by filtered cathodic vacuum arc, J. Phys. D. Appl. Phys. 37 (2004) 81. doi:10.1088/0022-3727/37/1/013.
[37]
M.A. Pimenta, G. Dresselhaus, M.S. Dresselhaus, L.G. Cançado, A. Jorio, R. Saito, Studying disorder in graphite-based systems by Raman spectroscopy, Phys. Chem. Chem. Phys. 9 (2007) 1276–1291. doi:10.1039/b613962k.
[38]
D. Mandal, I. Baker, Determination of the stored energy and recrystallization temperature as a function of depth after rolling of polycrystalline copper, Scr. Metall. Mater. 33 (1995) 645–650.
[39]
Y. Zhang, J.T. Wang, C. Cheng, J. Liu, Stored energy and recrystallization temperature in high purity copper after equal channel angular pressing, J. Mater. Sci. 43 (2008) 7326–7330. doi:10.1007/s10853-008-2903-8.
HIGHLIGHTS Novel cross-section images show a layered tribofilm and turbulent material flow. The amount of graphite in the composite only influences the friction slightly.
The friction is higher, and the wear rate lower, with current than without current. Oxides are not necessarily detrimental for the transfer of current. The presence of unoxidized metal is a requirement for a low contact resistance.
PII: DOI: Reference:
S0043-1648(17)31477-1 https://doi.org/10.1016/j.wear.2017.12.012 WEA102319
To appear in: Wear Received date: 16 October 2017 Revised date: 15 December 2017 Accepted date: 15 December 2017 Cite this article as: M. Grandin and U. Wiklund, Wear phenomena and tribofilm formation of copper/copper-graphite sliding electrical contact materials, Wear, https://doi.org/10.1016/j.wear.2017.12.012 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Wear phenomena and tribofilm formation of copper/copper-graphite sliding electrical contact materials M. Grandin*, U. Wiklund Tribomaterials group, Applied Materials Science, Uppsala University Ångström Laboratory, Box 534, SE-751 21 Uppsala, Sweden *Corresponding author: E-mail address: [email protected]; Tel.: +46 18 471 3069
Abstract Copper-graphite composites in reciprocating sliding against copper are investigated with regards to friction, wear and contact resistance. The tribological and electrical evaluation is complemented with surface analysis. It is shown that the presence of graphite in the composite greatly reduces the coefficient of friction and wear. However, the amount of graphite is not critical and only has a minor influence on friction and wear. It is observed that the coefficient of friction is slightly lower, while the wear rate is slightly higher, in pure mechanical tests than in tests with current. The contact resistance is greatly reduced by an increase in the copper content in the composite. Chemical analysis of the tribofilm that forms on the copper surface shows that it consists of graphite as well as Cu2O. It is shown that the mating couple with the highest amount of oxide in the tribofilm also has the lowest contact resistance. Hence, it is concluded that oxides are not necessarily detrimental for the contact resistance as long as there is unoxidized copper available. Novel cross-section techniques and images of the tribofilm contribute to a deeper understanding of how sliding electrical contact surfaces are affected by current and sliding motion. Keywords: Tribofilm, electrical contact, copper-graphite, contact resistance, friction
1. Introduction Continuously sliding electrical contacts are found, for example, where current or signals are transferred between one stationary and one rotating component. As such, they are tribological systems where friction and wear influence the performance. Carbon and graphite brushes are used for power transfer in everything from large generators to household appliances. Adding a metal, such as copper, will lower the resistivity and increase the current carrying capabilities [1]. Such metal-graphite composite materials are used to transfer power and signals in e.g. slip-ring assemblies and in locomotive pantograph strips for current collection. A satisfactory transfer of current, with low losses and low noise levels, are generally prioritized. Still, there is a lot to gain from improving also the tribological properties. In this paper, we present results on copper-graphite composite materials sliding against copper. Friction, wear and contact resistance are measured and special attention is paid to the contacting surfaces after testing, looking for material transfer, surface modifications and formation of tribofilms. Although the relationship between contact resistance and copper content in composite is rather straightforward, the correlation to friction and wear is more complicated. Both friction and wear benefit from small real contact areas, which can for example be achieved with hard materials. Although increasing the hardness usually has an impact on the friction and wear, it is important to remember that these parameters are not intrinsic to a specific material. For example, it is not necessarily the hardness of the materials that you start out with that makes a difference, but rather the tribofilm that is formed on the surfaces during sliding. It will be influenced by the specific circumstances during sliding, surrounding atmosphere, normal load etc. and its
nature is decisive for the friction, wear [2] and contact resistance. An optimal low friction situation is one where a thin tribofilm with low shear strength is formed and maintained on top of a hard surface. In this way, there will be benefits from both a small real contact area from the hardness and a low shear strength from the soft and easily sheared tribofilm. The formation of such tribofilms can be promoted by adding solid lubricants such as graphite, MoS2 or WS2. Due to their layered structure, these materials have low shear strength. In the case of graphite, however, the layered structure is necessary, but not sufficient. To achieve low friction, adsorption films are also necessary [3]. As system parameters, friction, wear and contact resistance need to be studied under specific circumstances. However, in the literature, focus is most often on either the electrical or the tribological properties. Transferring a current over a tribological interface may alter the friction and wear behaviour through increased temperature as the current is restricted to pass through the conducting spots and through a preferential direction of material transfer [4]. Still, copper-graphite materials are often studied without a transfer of current although electrical contacts are mentioned as a likely application [5–10]. Other studies apply current during testing, but do not measure the voltage drop [11–16] which hinders linkage between tribological and electrical performance to be studied. Some early work done on copper-graphite that is worth mentioning is that of J. Casstevens and co-workers [17–19] and Johnson and Moberly [20]. A lot of relevant knowledge was summarized by Rohatgi in 1992 [1] although his review includes other metal-graphites than coppergraphites as well as other applications than electrical contacts. More recent work focus on improvements related to different forms of graphite [5–8,21] or related to the manufacturing process [11,22,23], all of which have been evaluated against rotating steel. Correlations of the tribological properties and the amount of graphite in the composite show that increasing the vol.% of graphite up to 20% will lower the friction and wear, however, the positive effects on the tribological properties will not continue to increase with a further increase in the amount of graphite [1]. Scratch-testing of copper-graphite composites with varying amount of graphite up to 30 vol.% against a spherical diamond tip shows that the wear mechanism change from ploughing to microcutting with increasing load [24]. The same study also conclude that 6-20 vol.% of graphite is optimum for low shear forces, while more graphite result in particles coming loose, resulting in higher levels of wear. A few papers considering both tribological and electrical performance can be found in the literature, e.g. Bucca and Collina who make thorough investigations of the pantographcatenary system [25,26]. In addition, some studies of copper-graphite against copper or copperalloys can be mentioned; Cho et al. varied the amount of copper in the composite and the mechanical load and found that the voltage drop was minimized at 60 vol.%. and that the friction decrease with a decrease in copper content and mechanical load [27]. Shin and Lee found a correlation between brush wear and brush temperature [28]. The present work tries to increase the knowledge of the mechanisms behind such results as neither Bucca and Collina nor Cho, nor Shin and Lee investigate the contacting surfaces in detail.
2. Experimental 2.1. Test set-up and procedure Triboelectrical evaluation was performed in reciprocating sliding with two horizontal, crossed cylinders, at right angles, see Figure 1a. The voltage drop over the contact interface was measured using the four-point probes method where current-carrying electrodes are separated from voltagesensing electrodes avoiding voltage drop contributions in wires and other contact points than the one targeted.
Figure 1. a) Illustration of the crossed-cylinders and four-point probes configuration for voltage drop measurements. b) Test setup apparatus.
The cylinders are 20 mm long and 10 mm in diameter and the upper stationary cylinder was pressed against the lower moving cylinder using a spring applying 10 N normal load, see Figure 1b. A current controlled power supply was used to feed a direct current of 5 A through the contact, with the upper cylinder configured as cathode and the lower as anode. A stroke length of 5 mm and a 3 Hz frequency was used for 30 000 strokes (expect for the copper-copper reference test that was interrupted after 5000 strokes). Both the normal load and the friction force were measured during testing, as was the voltage drop from which the contact resistance was calculated assuming a constant current. Testing was performed without lubrication and at ambient lab conditions, the latter with variations in temperature and relative humidity small enough not to have any significant influence on the results.
2.2. Materials Copper cylinders were used in the stationary, cathode position. The copper cylinders are made from oxygen free copper with a purity of 99.95%. They were circumferentially polished prior to testing to an average Ra value of about 50 nm. The hardness of this material is 1.9 GPa and the elastic modulus is 140 GPa. As the moving, anode specimen, three different commercially available copper-graphite composite materials were used. These materials are pressed from powders and heat-treated. Some properties of these materials are listed in Table 1, as given by the manufacturer1. Note that the composite materials are named after the vol.% metal. The hardness is measured according to DIN IEC 413.303 (a method particularly developed for brushes, performed as a Rockwell test using a 10 mm ball and 60 kg load) and the resistivity is measured according to DIN IEC 413.402. After pressing, the composites had achieved a density that was 85% (Cu6), 95% (Cu26), and 75% (Cu70) of the maximum. The SEM images in Figure 2 give an indication of the distribution of metal and graphite in the composites. The composite nature of these specimen makes polishing challenging but all cylinders were circumferentially polished to a Ra value of approximately 200 nm. Nanoindentation hardness and elastic moduli were also determined separately for the two phases within the composites. On average, the graphite has HIT=0.6 GPa and EIT= 15 GPa and the copper HIT= 1.9 GPa and EIT=110 GPa. Although the hardness of copper is higher than the hardness of graphite, the softest of the three composites, according to the DIN IEC 413.303 method, is the one with the largest amount of copper. Table 1. Properties of the copper-graphite composites. Vol.% metal Hardness HR10/60 Resistivity (µΩm) Density (g/cm3)
Cu6 6 75 12.0 2.3
Cu26 26 85 0.82 3.2
Cu70 70 50 0.05 5.8
Figure 2. Polished copper-graphite composite surfaces illustrating the distribution of copper (bright) and graphite (dark).
2.3. Analysis The surfaces were analysed after testing with the objective of correlating the surface appearance with the tribological and electrical performance. White light interference profilometry for investigation of the worn composite surfaces was performed on an optical profiling system (WYKO NT1100, Veeco, USA). The length of the wear scar and the wear volume was also measured and used for calculating a specific wear rate. Surfaces were imaged in an SEM (Merlin, Zeiss,
1
Carbex AB, Sweden
Germany) and analysed with EDS (Aztec software). This SEM instrument is equipped with a FEG electron gun and X-Max Silicon drift detector. EDS compositions of the graphite wear scar (acquired using 5 kV acceleration voltage) and tribofilm on the copper cylinder (10 kV) are presented in at.%. The EDS composition outside the wear scar is also included to show the changes in composition. These compositions are of course numerically different from the vol.% compositions used during manufacturing and for naming the composites (c.p. Table 1). As EDS is not optimal for composition analysis of light elements, these results are mainly intended for qualitative comparisons. A broad ion beam (BIB) argon milling system (Ilion II, Gatan, USA) was used for milling cross-sections in the tribofilms formed on the copper cylinders. AGAR silver glue (thin silver flakes in methyl isobutyl ketone suspension) was used to mount the sample on a titanium blade for the milling operation. Raman spectroscopy was performed (inVia micro-Raman system, Renishaw, UK) using a 532 nm laser at a power below one mW. A nanoindenter (UNHT, CSM, Switzerland) was used for the hardness results presented above. A Berkovich tip was used and the Oliver & Pharr method [29]. For indentations in the copper phase in the copper-graphite the average load was 1450 µN and the depth 165 nm. For the graphite, the average load was 500 µN and the depth 190 nm. The results are in each case an average of at least 20 indentations
3. Results and discussion 3.1. Contact resistance Adding more metal to the copper-graphite composite results in a lower contact resistance, see Figure 3. This is an expected result as copper has a lower resistivity than graphite. Five identical tests were conducted and the standard deviation is shown as a shaded area around the mean curve. The contact resistance is increasing during the test period of 30 000 strokes, for Cu6 and most prominently for Cu70. From this observation it is clear that the contact situation changes over time; either the true contact area has been reduced or the materials in the contacting surfaces have been modified. The contact resistance for the three composite materials, approximated to a mean value over the 30 000 strokes, is plotted against the copper content in the inset in Figure 3. It turns out that, within this copper content interval, this relationship is roughly linear. It can be observed that the initial contact resistance behaviour differs greatly between Cu6 and Cu26. While Cu6 starts at a value much higher than its contact resistance after stabilization, Cu26 starts at a much lower value. In Cu6, which has a larger amount of graphite, some sliding is necessary to spread metal over a larger part of the contact surface and provide a low contact resistance. Initially, there are isolated spots of copper, which is not ideal for the transfer of current. For Cu26 there is enough copper to provide a low contact resistance already at the start, and the gradual increase is believed to be due to accumulation of insulating oxides at the interface.
Figure 3. The contact resistance increases gradually with sliding distance for Cu6 and Cu70, but it is almost constant for Cu26. The right inset shows how the contact resistance, averaged over the 30 000 strokes, decreases linearly with increasing vol.% of metal in the composite.
3.2. Coefficient of friction The coefficient of friction on the other hand does not relate to the amount of metal in the composite in a simple linear manner; see Figure 4 and Figure 5 As for the contact resistance, the standard deviation is shown as a shaded area around the mean curve. The friction is stable from around 15 000 strokes and hence a mean value of the last 15 000 strokes has been used to illustrate the mean steady state friction and its standard deviation in Figure 6. During the first 5000 strokes, a larger standard deviation, and hence large spread between tests, is observed, before they stabilize at very similar friction values. The inset in Figure 6 shows the friction of a copper-copper reference test, which makes it obvious that the presence of graphite is of great importance for the friction properties of this system. Among the composites tested in this study Cu26 results in the lowest coefficient of friction. Cu26 is the hardest of three composites, which could influence the friction provided that the shear strength in the interface is the same in all three cases. The coefficient of friction for the composite with the least amount of graphite has a mean value between the other two, but it can also be seen that the variation among identical tests is much greater for this composite. The reason for this is believed to be that the graphite is not evenly distributed in the copper matrix, which means that the amount of available graphite can differ from one test to another. For all three composites, the coefficient of friction is lower in tests without current than with current, although the high standard deviation for Cu70 makes these results uncertain. It can be estimated that the current induced heat in the contact spot dominates over friction induced heat. For Cu6 and Cu26 the voltage drop is above the softening voltage for copper (120 mV [30]), so it is likely that the additional heat due to current increases the size of the of a-spots. This will in turn increase also the real contact area and result in an increased friction.
Figure 4. The coefficient of friction is lower for Cu26 than for Cu6. For both composites, the friction is higher with than without current.
Figure 5. For the Cu70 composite, the variation between tests is larger than for the other two composites. A Cu-Cu reference test is shown in the inset.
Figure 6. Mean values of the steady state friction and its standard deviation for all three composites shows that Cu26 gives the lowest friction and Cu70 the highest standard deviation. For all composites, testing with current results in a slightly higher friction coefficient.
3.3. Effects on the composite cylinders In this section, results concerning the composite cylinder will be presented, first surface analysis of the wear scar and then the wear rate. The SEM images in Figure 7 give an idea of the appearance of the wear scar of the different composites. The difference in metal content in the composite is reflected also in the wear scar and hence the surfaces of the wear scars look very different from each other.
Figure 7. The wear scars in the different composites at two different magnifications (5 A).
There is no straightforward relationship between specific wear rate of the composite and its composition as can be seen in Figure 8. The variation within repeated tests is fairly high as indicated by the error bars; this is especially true for the Cu70 composite. Focussing on comparing Cu6 and Cu26, we see two trends; feeding 5 A current through the contact causes less wear of the composite than if no current is fed and more metal in the composite results in lower wear rate of the composite. The overall composition in the wear scar on the composite cylinders, tested with and without current, are shown in Figure 9, together with that of the untested composite for comparison. After testing, the Cu to C ratio has increased, i.e. there is an enrichment of metal as a result of sliding. This indicates that there is a preferential wear of graphite and/or that smearing of the malleable copper will cover part of the graphite. This is in accordance with the brighter shade inside than outside the wear scars in Figure 7. There is also an enrichment in oxygen after testing. Sliding will remove the oxidized surface layer, which uncover pure copper. This copper will, in turn, oxidize and with time, more and more oxide will accumulate in the wear track. In most cases, there appears to be more metallic copper in tests with than without current. The small additional heat in the contact, as a current is passing, may soften the copper slightly more and cause it to deform and cover the surface a little bit more, resulting in an enrichment of copper. Weakening of the graphite phase due to increased heat with current is one explanation of increased composite wear with current described in the literature [31]. Another possible mechanism could be that softening of the copper due to heat from the current, see section 3.2, will weaken the composite and result in wear. In our case, however, more heat is not associated with more wear. Instead, perhaps the observed increased amount of copper in the surface could counteract this weakening of the composite, resulting in a lower wear rate with than without current. For Cu6, the Cu to O ratio is below two, which indicates that all the copper in the surface is oxidized (assuming Cu2O, see the Raman results below).
Figure 8. The specific wear rates of the composite cylinders. The scatter between tests is large, particularly for Cu70.
Figure 9. Surface composition in the wear scar of the composite cylinders. Compositions of the composites, measured in the same way, are shown as reference. Sliding increases the relative amount of copper and oxygen.
3.4. Effects on the copper cylinders Tribofilms form on all copper cylinder surfaces, see Figure 10. For convenience, all images etc. are denoted by the counter copper-graphite cylinder composition in parentheses; (Cu6), (Cu26) and (Cu70), although it is the copper cylinder that is analysed. All tribofilms show a similar appearance although a slightly darker shade is seen for (Cu6) indicating a higher abundance of light elements like carbon and oxygen.
Figure 10. The tribofilms on the copper cylinder surface. The contrast is slightly different indicating a difference in composition. These examples are from tests at 5 A current but no differences are seen when comparing to tests without current.
From the EDS mapping shown in Figure 11 it is clear that for (Cu26) and particularly for (Cu6) there is an accumulation of carbon in the tribofilm which is not seen for (Cu70) where the amount of carbon on the surface is barely more than the natural carbon contamination always present on surfaces. The maps of copper also differ significantly. For (Cu6), and somewhat for (Cu26), the carbon and oxygen in the tribofilm is thick enough to shield the underlying copper. Mapping the oxygen concentration shows that the oxygen is more evenly distributed for (Cu70) than for the other two. The EDS composition of the tribofilm, as measured in the central part of each mark, is presented in Figure 12. As for the EDS analysis of the composite wear scar, for (Cu6) the Cu to O
ratio is below two, indicating that most of the copper is oxidized, while for (Cu70) metallic copper is present as well. Comparing the tribofilm composition with the wear scar composition, it is evident that the compositions are very similar on the two sides of the contact interface and there is arguably a tendency for higher Cu content with than without current.
Figure 11. EDS mapping of the tribofilms on the copper cylinder surface after testing at 5 A current shows that the composition seems to be related to the composition of the copper-graphite composite. It is clear that there is an enrichment of carbon and oxygen in the tribofilm.
Figure 12. Composition of the central region of the tribofilms. More graphite in the copper-graphite cylinder results in more graphite in the tribofilm.
As both graphite and copper-oxides are Raman active, this spectroscopy method can be used to detect the presence of graphite on the surface and to determine if oxides are present, and if so,
which kind they are. At lower wavenumbers five separate peaks can be identified in Figure 13: 114, 124, 148, 216, 528 and 627, and at least the last four of these are associated with Cu2O according to literature [32–36]. No other copper-oxides have peaks in the interval 110-120 cm-1 and the compound(s) giving rise to these peaks is unknown. The peaks above 1000 cm-1 in the Raman spectra can all be associated with disordered graphite [37]. At the very bottom of Figure 13, a Raman spectrum of the copper cylinder is shown and Cu2O is detected. Above that, spectra from the copper and graphite phases of the composite are also included as references where graphite is detected on both phases, and Cu2O only on the copper phase. Above that are spectra from three different tribofilms, produced against the different composites tested at 5 A current. In all cases, there is evidence for Cu2O and graphite. One spectrum from a test without current is also included to further prove the presence of oxide also in that case and to show that there is no significant difference between a test with and one without current. There are however, some differences depending on composition of the mating composite cylinder and this was investigated further by mapping with Raman. Figure 14 shows that the graphite layer covers more of the tribofilm surface with increasing amount of graphite in the mating composite cylinder. It also shows that the highest amount of surface oxide is in the (Cu70) tribofilm. Although the EDS analysis shows the most oxygen for (Cu70) and Raman mapping shows Cu2O in that tribofilm, the Cu70 gives the lowest contact resistance and it is concluded that presence of oxides is not detrimental for the electrical performance, as long as there is also metallic copper present that can transfer the current.
Figure 13. Raman spectroscopy of the tribofilms shows that Cu2O and graphite is present regardless of composite composition and weather being tested with or without current.
Figure 14. Raman spectroscopy mapping showing the distribution of graphite and Cu2O in the tribofilm.
The tribofilm was further analysed by cross-sectioning. Figure 15a illustrates how the cylinder is cut to produce the cross-section. Two reference surfaces are shown in Figure 15b and c. The first one is a cross-section of a copper cylinder that has not been subject to triboelectrical testing, the surface is smooth, and the copper grains are large. The second shows a cross-section of a wear mark from a Cu-Cu reference test that was run for 5000 strokes at 5 A current. Already after such a short running period, there is severe wear, an uneven surface and grain refinement to a depth of several micrometres. Figure 16 shows three examples of tribofilms formed against different coppergraphites in tests without current. Tribofilms formed at 5 A current are not shown but have a very similar appearance, no significant difference can be observed. For all three cases, we see an uneven surface and an uneven tribofilm, sometimes more than 5 µm thick. There is extensive deformation and mechanical mixing of materials with a tendency for vortex-like formations. There is also a grain refinement in the copper below the tribofilm. The deformation and strain, as a result of tribological action, will lower the recrystallization temperature of the copper surface [38,39]. This is the reason for recrystallization, and grain refinement, occuring already at ambient temperature. As this is observed also in tests without current, the heat from friction is apparently enough to give recrystallization and additional heat from the current is not necessary. The tribofilm has an overall darker contrast (more graphite) when there is more graphite in the mating composite, which is consistent with EDS results. As the graphite does not conduct current as well as copper, this difference can be expected to influence the contact resistance so that more graphite in the composite gives a higher contact resistance, which is consistent with the results shown in Figure 3. A closer look at the different components of a tribofilm is presented in Figure 17 where EDS has been used to analyse different spots in a cross-section of a (Cu70) tribofilm formed with 5 A current. The bottom part of the image, below the tribofilm, consists of pure unaffected Cu, although a tiny C peak resulting from surface contamination can be distinguished in the corresponding EDS spectrum. The second spectrum from the top is from a bright contrast area in the tribofilm, which consists mainly of copper, but tiny carbon and oxygen peaks are visible. The next spectrum below that is from a darker contrast area of the tribofilm, having larger amount of carbon and oxygen. A copper surface in air always has a thin natural oxide and this natural oxide is a part of all post-test analysis performed in this study. However, the fact that oxygen is observed throughout the depth of the tribofilm, makes it clear that the oxides observed in post-test analysis is not solely a thin natural oxide. Sometimes there are even darker contrast areas/particles, often found between tribofilm and copper. Apart from Cu and C, also Si and Al peaks are most often detected in these, together with a distinct oxygen peak. As this kind of dark features are neither present in the copper surface after polishing, c.p Figure 15b, nor in self-mated copper wear tracks, c.p Figure 15c, they must originate from the composites. Small amounts of Si and Al are present in the copper-graphite composites, most likely in the form of oxides. Being much harder than the copper, such oxides may wear the copper by abrasion before they become embedded. However, there are no signs of abrasive wear.
Figure 15. a) Illustration of how the cylinder is cut to produce a cross-section transverse to the sliding direction. b) A cross-section of a copper cylinder that has not been subjected to triboelectrical testing. b) A cross-section of a copper cylinder that has been tested for 5000 cycles against another copper cylinder.
Figure 16. Tribofilm cross-sections transverse to the sliding direction. The mating cylinders have different compositions, which is reflected in the tribofilms, but all have similar appearances with very uneven interfaces (10N load, without current).
Figure 17. EDS analysis of a cross-section transvers to the sliding direction showing that there is Si and Al present together with oxygen. The darker the contrast in the tribofilm, the more C and O is detected.
4. Conclusions Copper-graphite materials for sliding electrical contact applications have been evaluated in reciprocating sliding against copper. Testing has been performed both with and without current and the contact resistance is evaluated using the four-point probes method. The wear rate was measured and the surfaces analysed.
More graphite in the composite cylinder results in more graphite on the tribofilm forming on the copper cylinder. Both the composite cylinder surface and the tribofilm are enriched in copper during sliding and even more so if a current is fed through the contact. The tribofilm consists mainly of copper, graphite and Cu2O.
The contact resistance decreases linearly with increasing vol.% of metal in the composite.
The presence of oxides in the tribofilm is not detrimental for the contact resistance, which is shown for (Cu70) where the tribofilm contain the largest amount of oxygen, but the contact resistance is by far the lowest.
The coefficient of friction is similar for all three composites despite the large differences in graphite content. A total absence of graphite results in high friction and wear.
For all three composites, the coefficient of friction is higher, and the wear rate is lower, when a current is fed through the contact.
Increasing the amount of metal in the graphite composite decreases the wear rate.
Acknowledgements The authors would like to acknowledge Carbex AB for supplying materials.
References [1]
P.K. Rohatgi, S. Ray, Y. Liu, Tribological properties of metal-matrix-graphite particle composites, Int. Mater. Rev. 37 (1992) 129–149.
[2]
S. Jacobson, S. Hogmark, Tribofilms – On the crucial importance of tribologically induced surface modifications, in: G.K. Nikas (Ed.), Recent Dev. Wear Prev. Frict. Lubr., Research Signpost, 2010: pp. 197–225.
[3]
R.H. Savage, Graphite lubrication, J. Appl. Phys. 19 (1948) 1–10. doi:10.1063/1.1697867.
[4]
G.F. Dorsey, E.W. Glossbrenner, Sliding contacts for instrumentation and control, in: P.G. Slade (Ed.), Electr. Contacts, Princ. Appl., 2nd ed., CRC Press, Taylor & Francis group, 2014: pp. 1093–1149.
[5]
J. Wang, R. Zhang, J. Xu, C. Wu, P. Chen, Effect of the content of ball-milled expanded graphite on the bending and tribological properties of copper-graphite composites, Mater. Des. 47 (2013) 667–671. doi:10.1016/j.matdes.2013.01.008.
[6]
J. Li, L. Zhang, J. Xiao, K. Zhou, Sliding wear behavior of copper-based composites reinforced with graphene nanosheets and graphite, Trans. Nonferrous Met. Soc. China. 25 (2015) 3354–3362. doi:10.1016/S1003-6326(15)63970-X.
[7]
J. Kováčik, Š. Emmer, J. Bielek, L. Keleši, Effect of composition on friction coefficient of Cu-graphite composites, Wear. 265 (2008) 417–421. doi:10.1016/j.wear.2007.11.012.
[8]
S.F. Moustafa, S.A. El-Badry, A.M. Sanad, B. Kieback, Friction and wear of coppergraphite composites made with Cu-coated and uncoated graphite powders, Wear. 253 (2002) 699–710. doi:10.1016/S0043-1648(02)00038-8.
[9]
K. Rajkumar, S. Aravindan, Tribological performance of microwave-heat-treated coppergraphite composites, Tribol. Lett. 37 (2010) 131–139. doi:10.1007/s11249-009-9499-2.
[10]
W. Ma, J. Lu, Effect of sliding speed on surface modification and tribological behavior of copper-graphite composite, Tribol. Lett. 41 (2011) 363–370. doi:10.1007/s11249-010-9718-x.
[11]
M. Klement, O. Lott, A. Nagel, Synthesis and tribo-electric characterization of coppergraphite- composites with interpenetrating microstructure for sliding contacts, in: 27th Int. Conf. Electr. Contacts, Dresden, 2014: pp. 226–230.
[12]
H. Zhao, G.C. Barber, J. Liu, Friction and wear in high speed sliding with and without electrical current, Wear. 249 (2001) 409–414. doi:10.1016/S0043-1648(01)00545-2.
[13]
X.C. Ma, G.Q. He, D.H. He, C.S. Chen, Z.F. Hu, Sliding wear behavior of copper-graphite composite material for use in maglev transportation system, Wear. 265 (2008) 1087–1092. doi:10.1016/j.wear.2008.02.015.
[14]
G.X. Chen, F.X. Li, L. Dong, M.H. Zhu, Z.R. Zhou, Friction and wear behaviour of stainless steel rubbing against copper-impregnated metallized carbon, Tribol. Int. 42 (2009) 934–939. doi:10.1016/j.triboint.2008.12.011.
[15]
S. Du, F. Zhao, Y. Zhang, Friction and wear behavior of copper-graphite composite material in high-speed sliding with current, Adv. Mater. Res. 487 (2012) 411–415. doi:10.4028/www.scientific.net/AMR.487.411.
[16]
Y.A. Wang, J.X. Li, Y. Yan, L.J. Qiao, Effect of electrical current on tribological behavior of copper-impregnated metallized carbon against a Cu–Cr–Zr alloy, Tribol. Int. 50 (2012) 26–34. http://dx.doi.org/10.1016/j.triboint.2011.12.022.
[17]
J.M. Casstevens, H.G. Rylander, Z. Eliezer, Friction and Wear Characteristics, Wear. 49 (1978) 169–178.
[18]
J.M. Casstevens, H.G. Rylander, Z. Eliezer, Friction and wear properties of two types of copper-graphite brushes under severe sliding conditions, Wear. 50 (1978) 371–381.
[19]
J.M. Casstevens, H.G. Rylander, Z. Eliezer, Influence of high velocities and high current densities on the friction and wear behavior of copper-graphite brushes, Wear. 48 (1978) 121–130.
[20]
J. Johnson, L. Moberly, High-Current Brushes, Part I: Effect of Brush and Ring Materials, IEEE Trans. Components, Hybrids, Manuf. Technol. 1 (1978) 36–40.
[21]
K. Rajkumar, S. Aravindan, Tribological behavior of microwave processed coppernanographite composites, Tribol. Int. 57 (2013) 282–296. doi:10.1016/j.triboint.2012.06.023.
[22]
H. Sarmadi, A.H. Kokabi, S.M. Seyed Reihani, Friction and wear performance of coppergraphite surface composites fabricated by friction stir processing (FSP), Wear. 304 (2013) 1–12. doi:10.1016/j.wear.2013.04.023.
[23]
A. Guillet, E. Yama Nzoma, P. Pareige, A new processing technique for copper-graphite multifilamentary nanocomposite wire: Microstructures and electrical properties, J. Mater. Process. Technol. 182 (2007) 50–57. doi:10.1016/j.jmatprotec.2006.06.019.
[24]
J.K. Xiao, L. Zhang, K.C. Zhou, X.P. Wang, Microscratch behavior of copper-graphite composites, Tribol. Int. 57 (2013) 38–45. doi:10.1016/j.triboint.2012.07.004.
[25]
G. Bucca, A. Collina, A procedure for the wear prediction of collector strip and contact wire in pantograph-catenary system, Wear. 266 (2009) 46–59. doi:10.1016/j.wear.2008.05.006.
[26]
G. Bucca, A. Collina, Electromechanical interaction between carbon-based pantograph strip and copper contact wire: A heuristic wear model, Tribol. Int. 92 (2015) 47–56. doi:10.1016/j.triboint.2015.05.019.
[27]
K.H. Cho, U.S. Hong, K.S. Lee, H. Jang, Tribological properties and electrical signal transmission of copper-graphite composites, Tribol. Lett. 27 (2007) 301–306. doi:10.1007/s11249-007-9234-9.
[28]
W.-G. Shin, S.-H. Lee, An analysis of the main factors on the wear of brushes for automotive small brush-type DC motor, J. Mech. Sci. Technol. 24 (2010) 37–41. doi:10.1007/s12206-009-1135-4.
[29]
W.C. Oliver, G.M. Pharr, An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments, J. Mater. Res. 7 (1992) 1564–1583.
[30]
R. Holm, Electric Contacts; Theory and applications, 4th ed., Springer, Berlin, 1967.
[31]
Y.A. Wang, J.X. Li, Y. Yan, L.J. Qiao, Effect of pv factor on sliding friction and wear of copper-impregnated metallized carbon, Wear. 289 (2012) 119–123. doi:10.1016/j.wear.2012.04.006.
[32]
N. Wang, X. Cao, X. Cai, Y. Xu, L. Guo, Porous cuprite films: facile solution deposition and their application for nitrite sensing., Analyst. 135 (2010) 2106–10. doi:10.1039/c000086h.
[33]
Y. Wang, J. Ghanbaja, F. Soldera, S. Migot, P. Boulet, D. Horwat, et al., Tuning the structure and preferred orientation in reactively sputtered copper oxide thin films, Appl. Surf. Sci. 335 (2015) 85–91. doi:10.1016/j.apsusc.2015.02.028.
[34]
Y.S. Gong, C. Lee, C.K. Yang, Atomic force microscopy and Raman spectroscopy studies on the oxidation of Cu thin films, J. Appl. Phys. 77 (1995) 5422–5425. doi:10.1063/1.359234.
[35]
B. Balamurugan, B.R. Mehta, D.K. Avasthi, F. Singh, A.K. Arora, M. Rajalakshmi, et al., Modifying the nanocrystalline characteristics - Structure, size, and surface states of copper oxide thin films by high-energy heavy-ion irradiation, J. Appl. Phys. 92 (2002) 3304–3310. doi:10.1063/1.1499752.
[36]
Z.H. Gan, G.Q. Yu, B.K. Tay, C.M. Tan, Z.W. Zhao, Y.Q. Fu, Preparation and characterization of copper oxide thin films deposited by filtered cathodic vacuum arc, J. Phys. D. Appl. Phys. 37 (2004) 81. doi:10.1088/0022-3727/37/1/013.
[37]
M.A. Pimenta, G. Dresselhaus, M.S. Dresselhaus, L.G. Cançado, A. Jorio, R. Saito, Studying disorder in graphite-based systems by Raman spectroscopy, Phys. Chem. Chem. Phys. 9 (2007) 1276–1291. doi:10.1039/b613962k.
[38]
D. Mandal, I. Baker, Determination of the stored energy and recrystallization temperature as a function of depth after rolling of polycrystalline copper, Scr. Metall. Mater. 33 (1995) 645–650.
[39]
Y. Zhang, J.T. Wang, C. Cheng, J. Liu, Stored energy and recrystallization temperature in high purity copper after equal channel angular pressing, J. Mater. Sci. 43 (2008) 7326–7330. doi:10.1007/s10853-008-2903-8.
HIGHLIGHTS Novel cross-section images show a layered tribofilm and turbulent material flow. The amount of graphite in the composite only influences the friction slightly.
The friction is higher, and the wear rate lower, with current than without current. Oxides are not necessarily detrimental for the transfer of current. The presence of unoxidized metal is a requirement for a low contact resistance.