Friction and wear behavior of Al–CNT composites

Wear 307 (2013) 164–173 Contents lists available at ScienceDirect Wear journal homepage: www.elsevier.com/locate/wear Friction and wear behavior of Al–CNT composites Mina M.H. Bastwros a,c, Amal M.K. Esawi b,c,n, Abdalla Wifi a a b c Department of Mechanical Design and Production, Faculty of Engineering, Cairo University, Giza 12316, Egypt Department of Mechanical Engineering, The American University in Cairo, AUC Avenue, P.O. Box 74, New Cairo 11835, Egypt The Yousef Jameel Science and Technology Research Center, The American University in Cairo, AUC Avenue, P.O. Box 74, New Cairo 11835, Egypt art ic l e i nf o a b s t r a c t Article history: Received 18 February 2013 Received in revised form 14 August 2013 Accepted 17 August 2013 Available online 14 September 2013 Aluminum (Al)–carbon nanotube (CNT) composites are promising candidates for friction and wear applications. The wear behavior of Al–CNT composites, with up to 5 wt% homogeneously dispersed CNTs, is investigated in the present study and compared to that of pure aluminum processed using the same technique of cold compaction and hot extrusion. The effects of CNT content, sliding speed and applied load, on the wear behavior of the composites were studied. The morphologies of the wear surfaces were investigated using scanning electron microscopy (SEM). Hardness and wear resistance were found to increase significantly with CNT content. The wear rate of the 5 wt% CNT composite decreased by 78.8% compared to pure aluminum. This was accompanied by a decrease in the coefficient of friction. For samples with 5 wt% CNT, the wear rate and coefficient of friction were found to decrease with increasing sliding speed. The SEM investigation of the worn surfaces confirmed the dominant role played by the CNTs in enhancing the wear characteristics. CNTs were observed to be either partially or fully crushed forming a carbon film that covered the surface and acted as a solid lubricant enhancing the wear behavior significantly. & 2013 Elsevier B.V. All rights reserved. Keywords: Metal matrix composites Carbon Wear testing 1. Introduction Metal matrix composites are favored with superior properties compared to unreinforced metals such as higher specific strength, dimensional stability, higher elevated temperature stability and fatigue resistance. In addition, compared to polymeric matrix composites, they are favored with higher stiffness and strength, higher service temperature, higher thermal and electrical conductivity and better transverse properties [1]. Aluminum (Al) based metal matrix composites reinforced with ceramics whiskers, fibers and particles have been widely used in the automotive and aerospace applications which take advantage of their light weight as well as their high strength and wear resistance [2]. Recently, the superior mechanical and physical properties of carbon nanotubes (CNTs) have fascinated the research community. Although the main focus has been to use them to reinforce polymeric matrices, some trials were conducted to fabricate Al–CNT composites using different techniques. As with CNT–polymer composites, some challenges were faced; namely, the dispersion of the CNTs which are usually agglomerated and entangled together, and the poor wettability between the Al matrix and the CNTs. To-date, no studies have reported methods for successfully enhancing the bond between the CNTs and the Al matrix [3]. High energy ball milling is a promising technique to fabricate Al–CNT composites. Its effectiveness in dispersing the CNTs in the Al matrix has been reported in many studies. Moreover, it helps in increasing the interfacial bond between the Al and the CNTs by mechanical interlocking of the CNTs in the Al matrix [3–5]. In spite of its many advantages, ball milling has been reported by some researchers to result in damage and shortening of CNTs [5]. Most investigations have focused on the fabrication process and on studying the effect of adding the CNTs on the mechanical behavior of the produced composites [3,4,6–12]. A few studies have focused on investigating the wear behavior [13–16]. Such initial investigations are reporting promising results in that the CNTs are found to enhance the wear behavior of the composites, but – except for the study by Bakshi et al. [12] – mostly focus on using small amounts of CNTs. The current investigation aims to complement the previous efforts by providing an insight into the different wear mechanisms present in the Al–CNT composites, and in particular examines the wear behavior of individually dispersed CNTs at high CNT contents which have been difficult to produce in other studies. 2. Experimental procedure n Corresponding author at: Department of Mechanical Engineering, The American University in Cairo, AUC Avenue, P.O. Box 74, New Cairo 11835, Egypt. Tel.:þ 20 2 26153102; fax: þ 20 2 27952565. E-mail address: [email protected] (A.M.K. Esawi). 0043-1648/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.wear.2013.08.021 Air-atomized aluminum with a purity of 99.7% and an average particle size below 75 mm and CVD-MWCNTs with a purity of more than 95% and having an average diameter of 140 þ30 nm and an 165 M.M.H. Bastwros et al. / Wear 307 (2013) 164–173 average length of 7 72 mm were used in the present study. High energy ball milling, cold compaction and hot extrusion were used to fabricate both the pure aluminum as well as the composite samples. To isolate the effect of milling on the wear behavior of the aluminum matrix, unmilled pure aluminum samples were also prepared by cold compaction and hot extrusion and used for comparison. Ball milling was carried out at 400 rpm in a Retsch 400A planetary ball mill with a ball-to-powder weight ratio (BPR) of 5:1 using stainless steel milling balls having diameters of 10 mm. The low BPR was selected in order to limit any possible damage to the CNTs. The powder mix was milled for 30 min with a 10 min break intermitting every 10 min of milling to limit any heat build-up. Methanol was added as a process control agent (PCA). The amount used varied with the weight percentage of the CNTs in the charge, as suggested in another study by the authors [6]. 400 μl, 250 μl, 190 μl and 50 μl were added to the pure Al, Al–1 wt %, Al–2.5 wt%, Al–5 wt% CNTs, respectively. The ball milled powders were cold compacted in a steel die for 30 min using a uniaxial pressure of 475 MPa and then homogenized for 45 min at 500 1C before they were extruded into 10 mm diameter extrudates using an extrusion ratio of 4:1. The hardness of the pure Al and the Al–CNT composite samples was measured using a Vickers micro-hardness tester under an applied load of 10 N and a dwell time of 10 s. Ten readings were taken for each sample. The hardness of three samples was measured at each CNT content and the average was reported. The densities of the samples were obtained using a densitometer, based on the Archimedes principles, and the densities of three replicates at each condition were averaged and reported. Wear behavior tests were conducted on a pin-on-disk MultiAxis Tribometer, at room temperature and under dry sliding conditions. The produced samples were used as the pins which were tested by sliding them on a steel disk (EN31 high chrome tool steel). A sliding speed of 1.1 m/s and a normal applied load of 20 N were used for studying the effect of milling on the Al samples as well as the effect of the addition of the CNTs on the wear behavior. Additional sliding speeds of 0.18 m/s and 7.3 m/s and additional normal applied loads of 10 N and 15 N were used for the supplementary tests on the Al–5 wt% CNT samples which aimed at investigating the effect of the sliding speed and the applied load on the wear behavior. The coefficient of friction was recorded during sliding by means of a load transducer attached to the arm to which the pin was attached. The wear rate was calculated as the ratio between the mass loss and the sliding distance. An average of three replicate samples was taken as the wear rate at each condition. A LEO Supra 55 Field Emission Scanning Electron Microscope (FESEM) was used to examine the morphology of the worn surfaces after the wear test. X-ray diffraction analysis (XRD) of the wear debris was conducted using a Scintag XDS 2000 powder X-ray diffractometer with Cu Kα radiation to identify any compounds formed as a result of the wear test. 3. Results and discussion The experimentally measured densities of all samples are listed in Table 1. The theoretically estimated densities, calculated using the rule of mixture (Eq. (1)), are included for comparison purposes. The densities of pure Al unmilled and milled are also included. ρth ¼ ð1 V CNT ÞρAl þ V CNT ρCNT ð1Þ where ρth is the theoretical density of the composite, ρAl is the density of pure Al, ρCNT is the density of the MWCNT (1.9 g/cc), VCNT is the volume fraction of the CNTs in the composite. The relative density (the ratio between measured and calculated densities) of the pure Al milled and unmilled, and the Al–CNT Table 1 Density and average hardness of all samples. Pure Al unmilled Pure Al milled 1 wt CNT 2.5 wt CNT 5 wt CNT ρth (g/cc) ρexp (g/cc) ρexp/ρth (%) Hardness (HV) ( 7 SD) 2.7 2.7 2.69 2.67 2.64 2.668 2.664 2.647 2.633 2.618 98.8 98.6 98.4 98.6 99.1 39.4 7 1.49 74.6 7 2.09 78.17 1.97 84.5 7 3.06 95.2 7 2.18 composites were all above 98% which confirmed that the fabrication process produced well-consolidated samples with an acceptable percent of voids and porosities. Table 1 also presents the average Vickers micro-hardness values for all samples. It is noticeable that the hardness of the pure Al milled increased by 90% compared to the unmilled samples. This can be attributed to the strain-hardening effect of ball milling as reported in [6,7]. The addition of CNTs was observed to increase the hardness further to reach its maximum value of 95.2 HV for the Al–5 wt% CNT samples. It is argued that the hardness of the CNT composite samples increased because CNTs fill the micro-voids of the Al particles as noted in [15] as well as their reinforcement effect on the Al matrix [6]. 3.1. The effect of ball milling on the wear behavior of pure aluminum The coefficients of friction of both pure Al milled and unmilled samples were evaluated during sliding at a sliding speed of 1.1 m/s under an applied load of 20 N and were found to be similar. On the other hand, the wear rate of the pure Al milled was found to be 56.84 mg/km which is about 12% lower than that of the pure Al unmilled sample (64.42 mg/km). The enhancement in hardness is believed to have increased the load bearing capacity of the pure Al milled samples and consequently increased their wear resistance, as also noted in [17,18]. Fig. 1 shows SEM micrographs of representative worn surfaces of pure Al unmilled and milled and reveal distinct wear characteristics. As is clear in the images, the most dominant wear mechanisms are the abrasive and adhesive wear mechanisms, with the degree of severity noticeably lower in the case of the pure Al milled. Fig. 1(b) shows that the abrasion wear scars on the surface of the milled aluminum are with less plastic deformation on their banks, resulting in a smoother surface compared to the pure Al unmilled (Fig. 1(a)). The adhesion wear surface is also smoother with fewer features. The higher magnification micrographs (c) and (d) clearly show the difference in severity of plastic deformation between the two samples with the harder and less ductile milled surface showing noticeably fewer and shallower features. 3.2. Effect of CNT content on the wear behavior of the Al–CNT composites Fig. 2 shows the coefficient of friction of the pure Al milled and the Al–CNT composite samples, during sliding at a sliding speed of 1.1 m/s, under an applied load of 20 N. It is evident that the coefficient of friction decreased with the increase of the CNT content. Similar observations were reported by Choi et al. [13] who attributed the reduction in the coefficient of friction to the ultrafine grain structure of the aluminum matrix as well as the selflubricating effects of the CNTs. It is believed that with the increase in the CNT content, the fraction of the CNTs on the sample surface increases, and accordingly the direct contact between the Al matrix and the disk decreases, thus contributing to the reduction in the coefficient of friction. Both Kim et al. [15] and Jin-long et al. [16], who also investigated the wear behavior of CNT–Al 166 M.M.H. Bastwros et al. / Wear 307 (2013) 164–173 Fig. 1. SEM micrographs of the worn surfaces of (a) pure Al unmilled, (b) pure Al milled, (c) and (d) are magnified images of (a) and (b), respectively. 0.8 0.7 0.6 Pure Al milled 0.5 1wt% CNT 0.4 56.84 54.4 60 Wear rate (mg/km) Coefficient of Friction 0.9 Sliding speed: 1.1m/s Applied load : 20 N 50 40 30 18.1 20 12.04 10 0.3 2.5wt% CNT 0.2 0 Pure Al milled 1 wt% CNT 2.5 wt% CNT 5 wt% CNT 5wt% CNT 0.1 0 70 Sliding speed: 1.1 m/s Applied Load: 20 N 1 Fig. 3. Wear rate versus CNT wt%. 0 500 1000 1500 Sliding Time (sec) Fig. 2. Plot of the coefficient of friction versus the sliding time for pure Al milled as well as the Al–CNT samples. composites, reported that a film of carbon can cover the wear surface and act as a solid lubricant that decreases the coefficient of friction. The formation of this carbon film in the present work is believed to have also resulted in a more smooth friction trace, as is evident upon comparing the coefficient of friction traces of all the samples. It is worth noting that the pure Al milled sample showed a sharp decrease in the coefficient of friction at approximately 1500 s which can be attributed to the temperature increase during sliding, as will be discussed later, which resulted in softening of the sample. On the other hand, a similar drop did not occur in the case of the Al–CNT composites. It is thought that this is due to the following reasons: the presence of the CNTs helped to retain the strength and hardness of the Al–CNT composites at higher temperatures as observed in other studies by the authors [7], the decrease in the coefficient of friction helped to reduce the heat generated during sliding, and the outstanding thermal conductivity of the CNTs increased the thermal conductivity of the Al–CNT composites and thus helped to liberate the heat generated. An estimation of the temperature increase due to frictional heat is presented in the discussion section. Fig. 3 compares the wear rate of the Al–CNT composites with different weight percentages of CNT. It is noted that upon the addition of 1 wt% CNT the wear rate decreased modestly compared to that of the pure aluminum milled. However, as the CNT content was increased further, the wear rate was observed to decrease significantly to reach 78.8% reduction for the 5 wt% CNT sample. This considerable decrease in the wear rate with the increase in CNT content is attributed to the strengthening of the samples by the CNT reinforcements as reported in other studies by the authors [7,8], the significant increase in hardness as presented in Table 1 as well as the decrease in the coefficient of friction, as shown in Fig. 2. The steady decrease in wear rate with increase in CNT content also confirms the uniform dispersion of the CNTs as a result of the effective ball milling technique employed in the present work. Other researchers reported a decrease in wear rate with the addition of CNT but then reported an increase as the CNT amount increased and attributed this to CNT agglomeration which resulted in pores, poor densification and the consequent separation of conglomerated particles which weakly bond to the surface [13,15–17]. Fig. 4 shows SEM micrographs for representative worn surfaces of pure Al milled and Al–5 wt% CNT samples representing the two extreme conditions. As noted earlier, in the case of the pure Al milled, the dominant wear mechanisms are the adhesion and abrasion wear mechanisms. Upon adding the CNTs to the Al matrix, the dominant wear mechanism changed to being largely abrasion wear. Closer examination reveals that the wear tracks are M.M.H. Bastwros et al. / Wear 307 (2013) 164–173 167 Fig. 4. SEM micrographs of the worn surfaces of (a) pure Al milled, (b) Al–5 wt% CNT, (c) and (d) are magnified images of (a) and (b), respectively. featureless and smooth along the sliding direction, as also observed in [13]. The CNT addition heavily reduced the severity of the adhesion wear compared to the pure Al milled, as shown in Fig. 4(b) and (d). It can also be noticed in the higher magnification micrographs in Fig. 4(c) and (d) that the surface roughness of the pure Al milled is higher with large plastic deformations and evidence of delamination and microploughing that are not observed in the Al–5wt% CNT sample. Fig. 5 shows SEM micrographs for representative worn surfaces of Al–1 wt% CNT, Al–2.5 wt% CNT and Al–5 wt% CNT samples and capture the effect of increasing the CNT content on the characteristics of the worn surfaces. Analysis of the worn surface of the Al– 1 wt% sample reveals less severe scars and material delamination compared to the pure milled sample. A clear change in the wear mechanism is observed as the CNT content is increased to 2.5 wt%. The dominant wear mechanism changed from adhesion (in the 1 wt% CNT) to abrasion (2.5 wt% and 5 wt% CNT), generating smoother surfaces without any flake-like wear scars on the worn surface which are typical characteristics of adhesive wear. As noted earlier, as the CNT wt% increased in the Al–CNT composites, the hardness of the composite increased which limited the plastic deformation of the surface. In addition, the increase in the fraction of CNTs on the surface reduced the direct contact between the aluminum matrix and the steel disk and reduced the coefficient of friction due to the CNTs self-lubricating properties [16]. 3.3. Effect of normal applied load on the wear behavior of the Al–CNT composites Fig. 6 presents the coefficient of friction as a function of sliding time for the Al–5 wt% CNT composite under different applied loads. The average coefficient of friction is observed to decrease with the increase of the applied loads. This can be attributed to the increase in the ploughing force (friction force) and the penetration inside the sample, as noted in [19], which in turn increases the amount of CNTs pulled out of the matrix, or crushed, and thus increases the efficiency of lubrication. It is also noted that the fluctuation in the coefficient of friction is less at lower loads. This can be related to the nature of the CNT deformation under the wear tester, as is presented in a subsequent section. It is believed that the CNTs are wearing off gradually – rather than being crushed – under this lower load which is causing a more smooth friction behavior. Moreover, the coefficient of friction is observed to decrease and then increase with sliding time when using a 15 N load. This is thought to be caused by possible surface damage at this higher load, as noted by Choi et al. [13]. The same trend was observed to occur with the 20 N load. However, in the case of the 20 N load, the coefficient of friction was observed to decrease again. At this high load, the contact area temperature is expected to rise more severely which leads to softening and accordingly lower friction. Fig. 7 shows the wear rate of the Al–5wt% CNT composite under the different applied loads. The wear rate increased with the increase of the applied load due to the increased penetration which increased the amount of material ploughed. The negative impact of CNTs on the wear resistance was attributed to excessive sub-surface fracturing and delamination at high loads (above 15 N) by Al-Qutub et al. [20] for Al6061–1 wt% CNT composites. The same trend has also been reported in [13,16] but for samples with up to 3% CNT. Fig. 8 shows the SEM micrographs of the worn surfaces of Al–5 wt% CNT samples tested under different applied loads. It can be noticed that the width of the abrasion scars increased with the increase of the applied load which supports the results of the wear rate test presented in Fig. 7. Moreover, it can be seen in Fig. 8(a and b) that the surface tested at the highest applied load is clear, smooth and with minimal features, and that with the decrease of load the surface appears slightly rougher (Fig. 8(e and f)). This observation supports the earlier argument that under the lower load, the CNTs are not fully crushed but are gradually wearing off. 3.4. Effect of changing the sliding speed on the wear behavior of the Al–CNT composites Fig. 9 presents the coefficient of friction as a function of sliding time for the Al–5 wt% CNT composite under different sliding speeds. It is noticed that the coefficient of friction is lower at the higher speed which can be attributed to the softening of both the pin and the disk during sliding, as noted in [13,21] since at higher sliding speeds, the frictional heat is higher resulting in a higher 168 M.M.H. Bastwros et al. / Wear 307 (2013) 164–173 Fig. 5. SEM micrographs of representative worn surfaces of Al–CNT samples: (a) 1 wt%, (c) 2.5 wt%, (e) 5 wt%, (b), (d) and (f) are magnified images of (a), (c) and (e), respectively. 12.04 10.86 12 0.5 0.4 10 N 15 N 20 N 0.3 0.2 0.1 0 sliding speed: 1.1 m/s 14 0.6 0 500 1000 1500 Sliding Time (sec) Fig. 6. Plot of the coefficient of friction as a function of sliding time for the Al–5 wt% CNT composite under different applied loads. Wear rate (mg/km) Coefficient of Friction 0.7 9.68 10 8 6 4 2 0 10N 15N 20N Fig. 7. Wear rate of the Al–5 wt% CNT composite under different applied loads. flash temperature at the contact points [22], as will be elaborated in the discussion section. Fig. 10 shows the wear rate of Al–5 wt% CNT composite tested under different sliding speeds. The wear rate is observed to decrease with the increase in the sliding speed. A similar trend was reported by Choi et al. [13]. Although increasing the sliding speed is expected to result in a dramatic increase in surface temperature and accordingly more severe wear [22,23], the observed decrease in wear rate with increasing sliding speed suggests that the presence of the CNTs in the samples and their contribution to lowering the frictional heat is playing a more dominant role. It is also suggested that as the sliding speed increased, the higher centrifugal force acting on the disk threw the wear debris away from the disk, reducing the rubbing abrasion medium and thus contributing to the decrease in the wear rate. SEM observations of the worn surfaces (Fig. 11) confirmed the above especially the higher magnification images (Fig. 11b, d and f) where it can be noticed that the amount of the wear debris present on the surface decreased as the sliding speed increased. The figures show a large amount of wear debris on the surface tested at 0.18 m/s whereas the surface of the 7.3 m/s sample is nearly free from wear debris. It can also be noticed from Fig. 11(a, c and e), that the surface roughness of the samples is decreasing 169 M.M.H. Bastwros et al. / Wear 307 (2013) 164–173 Fig. 8. SEM micrographs of the Al–5 wt% CNT worn surface under different applied loads: (a) 20 N, (c) 15 N, (e), 10 N, (b), (d) and (f) are magnified images of (a), (c) and (e), respectively. 25 1.1m/s 0.18m/s 0.6 Wear rate (mg/km) Coefficient of Friction 0.7 0.5 0.4 0.3 0.2 applied load : 20 N 22.23 20 15 12.04 10.98 10 5 0.1 0 0 0 500 1000 1500 Sliding Time (sec) Fig. 9. Coefficient of friction versus the sliding time of the Al–5 wt% CNT composite at different sliding speeds. with the increase of the sliding speed, which suggests that the higher sliding speed generates a smoother surface. 3.5. Discussion Although reference was made in the few published studies investigating the tribological properties of Al–CNT composites to the low friction coefficient and accordingly the self-lubricating role of CNTs in enhancing the wear behavior [13,15,16] in addition to similar observations reported in other studies investigating the role of CNTs in enhancing the wear behavior of ceramic 0.18m/s 1.1m/s 7.3 m/s Fig. 10. Wear rate of the Al–5 wt% CNT Al composite at different sliding speeds. CNT composites [17,24–27], no previous study attempted to explain the observed wear behavior by closely investigating the behavior of the individual CNTs as they are subjected to the wear tester. It is believed that due to the severe wear conditions, the aluminum matrix is gradually abraded exposing the CNTs which are dispersed and embedded within it. Fig. 12 presents an SEM image of the worn surface of one of the samples showing two exposed CNTs. The abrasion of the aluminum matrix to expose the individual CNTs observed in the present study is, however, different from that reported by Balani et al. [26] in which samples of Al2O3 with a high content of CNTs are reported to have increased densification which prevented the protrusion of CNT ends and reduced material loss by abrasion. As is clear in Fig. 12, the exposed CNTs remained attached to the aluminum matrix and 170 M.M.H. Bastwros et al. / Wear 307 (2013) 164–173 Fig. 11. SEM micrographs of the Al–5 wt% CNT worn surface at different sliding speeds, (a) 0.18 m/s, (c) 1.1 m/s, (e) 7.3 m/s, (b), (d), (f) are magnified images of (a), (c) and (e), respectively. Fig. 12. Attached CNTs on the worn surface of the Al–5 wt% CNT sample. appear intact in spite of being partially exposed to the wear tester. However, upon further examination at higher magnifications, as in Fig. 13, it becomes apparent that some CNTs are partially crushed, or worn out with possibly some parts changing to amorphous carbon. For example, in Fig. 13(a), a damaged CNT can be seen with some of its outer walls already abraded and in Fig. 13(b), an attached CNT has one worn end with a reduced diameter. Both crushing and gradual wear of the CNTs are assumed to have taken place during abrasion wear which, as revealed by the SEM results reported earlier, is the dominant wear mechanism in most Al–CNT composite samples. Whether crushing or gradual wear occurred is thought to depend on the applied load, as noted earlier. It is also believed that a carbon film is formed from the debris of the CNTs (small parts of the graphene sheets that make up the CNTs walls). According to Scharf et al. [28] who also observed the formation of a thin graphitic film in the case of Ni–CNT composites, the low Fig. 13. Attached CNTs on Al–CNT composites worn surfaces, (a) 2.5 wt% CNT, (b) 5 wt% CNT. interfacial shear strength of this graphitic film is responsible for lowering the friction coefficient. Similarly, Dong et al. [29] reported that a carbon film was formed when testing Cu–CNT composites and is believed to have inhibited oxidation and wear. M.M.H. Bastwros et al. / Wear 307 (2013) 164–173 The crushing of the CNTs and accordingly the formation of a lubricating film that severely reduces the coefficient of friction was also recently reported by Puchy et al. [24] who studied the wear resistance of Al2O3–CNT ceramic nanocomposites. Similarly, Balani et al. [25] reported enhanced lubrication by graphitized CNTs in Al2O3–CNT nanocomposite coatings and Lahiri et al. [27] reported a reduced coefficient of friction due to the lubrication effect of delaminated graphene layers in CNT-reinforced hydroxyapatite. The formation of a carbonaceous film was also reported by Rajkumar and Aravindan [30], Lin et al. [31], Perez-Bustamante et al. [32] and believed to result in a lower coefficient of friction and a lower wear rate in copper–CNT and Al2024–CNT composites. It is suggested that the formation of the lubricating carbon film in the present study reduced the direct contact between the sample and the disk and reduced the wear rate significantly – especially as the film volume increased with the increase in the CNT content – as confirmed by the wear rate and friction coefficient results presented earlier. In addition to the attached but abraded CNTs, SEM observations revealed (Fig. 14) some CNTs resting freely on the worn surface of an Al–5 wt% CNT composite sample after apparently being completely detached from the aluminum matrix. It can be noticed that the CNTs are shorter than the as-received ones and that the ends of the CNTs are damaged and worn out. As they rested freely on the surface, the CNTs were either partially or totally crushed, or were gradually worn out; thus contributing Fig. 14. Detached CNTs on the worn surfaces of Al–5 wt% CNT composites. 171 to the carbon film. Due to their high hardness and small size, as the detached CNTs rolled between the surfaces, they may have played a part in a “three-body abrasion” mechanism by contributing towards abrading the two mating surfaces (aluminum matrix and the counterpart steel disk) slightly and reducing the wear rate [16] and the friction coefficient [17]. Three body abrasion causing minimal wear was reported in [15]. The wear debris of the tested samples had the black color of carbon. Fig. 15 shows a photo of the wear track (Fig. 15(a)) after the wear test with the wear debris on its banks as well as a photo of the wear debris (Fig. 15(b)) that was collected on a white sheet of paper. The dominant black color verifies that the aluminum debris were covered with the carbon film which also stained the white paper. As suggested in [15,16,28–32], it is concluded that the carbon film that formed during sliding acted as a solid lubricant, and had a major impact on the attained wear properties. Analysis of the wear debris using the SEM (Fig. 15(c)) revealed that it is made up of fragments of CNTs thus confirming the earlier assumption that both loose and detached CNTs have been crushed or gradually worn out and have contributed to the formation of the lubricating carbon film. Bakshi et al. [12] found through Raman spectroscopy analysis of their wear debris that significant amorphization and highly defective carbon is present in the wear debris which corroborates the CNT damage reported in the current study. These findings are however contrary to those reported by An et al. [17] who investigated the tribological properties of hot-pressed alumina CNTcomposites and reported that the CNTs maintained their shape after the wear test. They also argued that the decrease in friction coefficient with increasing CNT content is due to the rolling motion of the CNTs at the sliding surface. XRD analysis was conducted on the debris to explore the possibility of tribochemical reactions (more specifically carbides and oxides) that may be affecting the attained wear rates. In addition, XRD was used to analyze the effect of wear on the CNTs. Fig. 16 shows the XRD pattern of the wear debris. It confirms that the debris is aluminum with traces of Al4C3. No (002) peak at around 261 (corresponding to the inter-planar spacing of 0.34 nm of CNTs was observed). Previous research has indicated that the (002) peak disappears upon dispersion of the CNTs in the matrix [6,8] which confirms that there are no CNT agglomerates in the sample. The XRD analysis also confirms that the carbon film has inhibited oxidation wear especially as its volume increased with the increase in the CNT content in the samples. Bakshi et al. [12] reported increased oxidative wear in their Al–Si–CNT samples and attributed it to the lack of carbon film formation. The flash temperature model presented by Ashby et al. [22] was used to calculate the local or flash temperature at the asperities or points of contact between the two contacting surfaces; in this case the steel disk and the aluminum–CNT pin. The following values were obtained for the pure, 1, 2.5 and 5 wt% composite samples respectively: 200, 187, 117, and 89 1C when using an applied load of Fig. 15. (a) Wear track after the wear test with the wear debris on its banks, (b) wear debris after it was collected on a white sheet of paper and (c) SEM image showing the morphology of the wear debris. 172 M.M.H. Bastwros et al. / Wear 307 (2013) 164–173 100 90 80 Aluminium carbide (Al4C3)peaks Intensity 70 Al (111) 60 50 Al (200) 40 Al (220) 30 Al (311) Al (222) 20 10 0 10 20 30 40 50 60 70 80 90 100 Angle 2-theta, degrees Fig. 16. XRD spectrum of the debris of an Al–5 wt% sample showing some aluminum carbide peaks. 20 N and a sliding speed of 1.1 m/s. Additionally the flash temperature for the 5 wt% sample tested at a sliding speed of 7.3 m/s under a 20 N load was found to be 593 1C confirming the dramatic increase in flash temperature with the increase in sliding speed [23]. Due to the special nature of the material used in the current study – the CNTs – it can, however, be argued that the model has overestimated the flash temperature for the samples with 2.5 and 5 wt% CNT. The reason is that the model calculates the true contact area based on the load, the hardness and the coefficient of friction. The current investigation is nonetheless revealing as discussed earlier that the true contact area is increasing with the increase in CNT content as the CNT film spreads over the surface. Taking this into consideration and fine tuning the model accordingly gives flash temperatures of 96, 60 and 245 oC for the 2.5 wt%, 5 wt% (v ¼1.1 m/s) and 5 wt% (v¼ 7.3 m/s). The estimated flash temperatures seem to be in line with observations made during the wear test about the temperature rise in the samples. The following assumptions were, however, made: (1) the thermal conductivity of the MWCNTs is 3000 W/mK, (2) because of varying reports in the literature on the exact value of the thermal conductivity of MWCNTs when embedded in matrices, the thermal conductivities of the Al–CNT composites are assumed to follow the rule-of-mixture with a CNT efficiency of 50% to account for the fact that the CNTs are not aligned in the longitudinal direction, CNT wall defects, and Al–CNT interfacial contact resistance, (3) no thermal contact resistance between the pin and the heat sink (the steel chuck holding the pin) and so using the physical lengths of the pin and disk in calculations. In the case of our composites, the calculated temperatures confirm the role played by the carbon film which increases in area with the CNT content thus covering the pin surface and leading to the reduction in the flash temperature. Examination of the temperature map presented by Lim and Ashby [23] reveals that the flash temperature depends strongly on velocity and much less so on load. The maximum flash temperature calculated using Ashby's model – which is 245 1C for the 5% sample tested at the highest sliding speed and the highest load – is still lower than the temperatures reported in the literature to result in the formation of aluminum carbide. The traces of Al4C3 observed during the XRD analysis are believed to have formed during the composite fabrication process, as reported in other studies by the authors [7,8]. 4. Conclusions In this work, Al–CNT composites were fabricated by mixing the aluminum particles and the MWCNTs using high energy ball milling, followed by cold compaction of the mix and finally hot extrusion. The CNT content was varied from 0 wt% up to 5 wt% which resulted in a significant increase in the hardness and the wear resistance, as well as a decrease in the coefficient of friction. The Al–5 wt% CNT composite was tested at different sliding speeds and under different applied loads. On increasing the sliding speed, the wear rate and the coefficient of friction were observed to decrease; but on increasing the applied load, the wear rate increased while the coefficient of friction decreased. Although the wear behavior of the Al–CNT composites is affected by the hardness and the CNT content in the composite, it is argued that the CNT content is the dominant factor affecting the wear behavior of the Al–CNT composites. It is also concluded that a homogenous CNT distribution – coupled with good densification and enhanced hardness – is essential for enhanced wear behavior at high CNT contents. SEM observations showed that the CNTs, whether still attached or became loose from the Al matrix, were crushed or worn out to form a carbon film that acted as a solid lubricant and reduced the wear rate and the coefficient of friction. At the same time, owing to the self lubrication properties of CNTs the exposed CNTs which remained attached to the matrix reduced the surface fraction of the worn surface in direct contact with the rubbing surface, and consequently reduced the wear rate and the coefficient of friction. The improved wear behavior reported here coupled with the strengthening effect of the CNTs, makes Al–CNT composites excellent candidates for applications subjected to wear and relatively higher strength. Acknowledgments The authors acknowledge the financial support by the Yousef Jameel Science and Technology Research Center at the American University in Cairo, Egypt as well as the Tribology and Spare Parts Center at the Faculty of Engineering, Cairo University for the use of the facilities. Grateful thanks to Prof. Mahmoud El-Sherbiny for many helpful discussions. The authors also wish to acknowledge the technical assistance of Mr. Rami Wasfi, Mr. Gebreel and Mr. Mohamed Gebreel. References [1] N. Chawla, K. Chawla, Metal Matrix Composites, first ed., Springer Science, New York, 2006. [2] T. Miyajima, Y. Iwai, Effect of reinforcement on sliding wear behavior of aluminum matrix composites, Wear 225 (2003) 606–616. [3] A. Esawi, K. Morsi, Dispersion of carbon nanotubes (CNTs) in aluminum powder, Composites Part A: Applied Science 38 (2007) 646–650. [4] R. George, K.T. Kashyap, R. Rahul, S. Yamdagni, Strengthening in carbon nanotube/aluminium (CNT/Al) composites, Scripta Materialia 53 (2005) 1159–1163. [5] D. Poirier, R. Gauvin, R. Drew, Structural characterization of a mechanically milled carbon nanotube/aluminum mixture, Composites Part A: Applied Science 40 (2009) 1482–1489. [6] A.M.K. Esawi, K. Morsi, A. Sayed, M. Taher, S. Lanka, Effect of carbon nanotube (CNT) content on the mechanical properties of CNT-reinforced aluminium composites, Composites Science and Technology 70 (2010) 2237–2241. [7] A.M.K. Esawi, K. Morsi, A. Sayed, A. Abdel Gawad, P. Borah, Fabrication and properties of dispersed carbon nanotube–aluminium composites, Materials Science and Engineering A: Structural Materials 508 (2009) 167–173. [8] A.M.K. Esawi, K. Morsi, A. Sayed, M. Taher, S. Lanka, The influence of carbon nanotube (CNT) morphology and diameter on the processing and properties of CNT-reinforced aluminium composites, Composites Part A: Applied Science 42 (2011) 234–243. [9] H.J. Choi, G.B. Kwon, G.Y. Lee, D.H. Bae, Reinforcement with carbon nanotubes in aluminum matrix composites, Scripta Materialia 59 (2009) 360–363. [10] K. Morsi, A. Esawi, Effect of mechanical alloying time and carbon nanotube (CNTs)content on the evolution of aluminum (Al)–CNT composite powder, Journal of Materials Science 42 (2007) 4954–4959. [11] D.S. Lim, D.H. You, H.J. Choi, S.H. Lim, H. Jang, Effect of CNT distribution on tribological behavior of alumina–CNT composites, Wear 259 (2005) 539–544. [12] S.R. Bakshi, A.K. Keshri, A. Agarwal, A comparison of mechanical and wear properties of plasma sprayed carbon nanotube reinforced aluminum M.M.H. Bastwros et al. / Wear 307 (2013) 164–173 [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] composites at nano and macro scale, Materials Science and Engineering A: Structural Materials 528 (2011) 3375–3384. H.J. Choi, S.M. Lee, D.H. Bae, Wear characteristic of aluminum-based composites containing multi-walled carbon nanotubes, Wear 270 (2010) 12–18. H. Mindivan, E. Sabri Kayali, H. Cimenoglu, Tribological behavior of squeeze cast aluminum matrix composites, Wear 265 (2008) 645–654. I. Kim, J. Lee, G. Lee, S. Baik, Y. Kim, Y. Lee, Friction and wear characteristics of the carbon nanotube–aluminum composites with different manufacturing conditions, Wear 267 (2009) 593–598. J. Jin-long, W. Hai-zhongY. HuaX. Jin-cheng, Fabrication and wear behavior of CNT/Al composites, Transactions of Nonferrous Metals Society of China 17 (2007) 113–116. J.-W. An, D.-H. You, D.-S. Lim, Tribological properties of hot-pressed alumina– CNT composites, Wear 255 (2003) 677–681. M.D. Bermudez, G. Martinez-Nicolas, F.J. Carrion, I. Martinez-Mateo, J. A. Rodriguez, E.J. Herrera, Dry and lubricated wear resistance of mechanicallyalloyed aluminum-base sintered composites, Wear 248 (2001) 178–186. P. Blau, ASM Handbook: Friction, Lubrication and Wear TechnologyASM International, OH (1992). A.M. Al-Qutub, A. Khalil, N. Saheb, A.S. Hakeem, Wear and friction behavior of Al6061 alloy reinforced with carbon nanotubes, Wear 297 (2013) 752–761. H.J. Kim, A. Emge, S. Karthikeyan, D.A. Rigney, Effect of tribooxidation on sliding behavior of aluminum, Wear 259 (2005) 501–505. M.F. Ashby, J. Abulawi, H.S. Kong, Temperature maps for frictional heating in dry sliding, Tribology Transactions 34 (1991) 577–587. S.C. Lim, M.F. Ashby, Overview no. 55 wear-mechanism maps, Acta Metallurgica et Materialia 35 (1987) 1–24. 173 [24] V. Puchy, P. Hvizdos, J. Dusza, F. Kovac, F. Inam, M.J. Reece, Wear resistance of Al2O3–CNT ceramic nanocomposites at room and high temperatures, Ceramics International 39 (2013) 5821–5826. [25] K. Balani, S.P. Harimkar, A. Keshri, Y. Chen, N.B. Dahotre, A. Agarwal, Multiscale wear of plasma-sprayed carbon-nanotube-reinforced aluminum oxide nanocomposite coating, Acta Materialia 56 (2008) 5984–5994. [26] K. Balani, Y. Chen, S.P. Harimkar, N.B. Dahotre, A. Agarwal, Tribological behavior of plasma-sprayed carbon nanotube-reinforced hydroxyapatite coating in physiological solution, Acta Biomaterialia 3 (2007) 944–951. [27] D. Lahiri, V. Singh, A.K. Keshri, S. Seal, A. Agarwal, Carbon nanotube toughened hydroxyapatite by spark plasma sintering: microstructural evolution and multiscale tribological properties, Carbon 48 (2010) 3103–3120. [28] T.W. Scharf, A. Neira, J.Y. Hwang, J. Tiley, R. Banerjee, Self-lubricating carbon nanotube reinforced nickel matrix composites, Journal of Applied Physics 106 (2009) 013508-1–013508-7. [29] S.R. Dong, J.P. Tu, X.B. Zhang, An investigation of the sliding wear behavior of Cu-matrix composite reinforced by carbon nanotubes, Materials Science and Engineering A: Structural Materials 313 (2001) 83–87. [30] K. Rajkumar, S. Aravindan, Tribological studies on microwave sintered copper– carbon nanotube composites, Wear 270 (2011) 613–621. [31] C.B. Lin, Z.-C. Chang, Y.H. Tung, Y.-Y. Ko, Manufacturing and tribological properties of copper matrix/carbon nanotubes composites, Wear 270 (2011) 382–394. [32] R. Pérez-Bustamante, J.L. Bueno-Escobedo, J. Jiménez-Lobato, I. Estrada-Guel, M. Miki-Yoshida, L. Licea-Jiménez, R. Martínez-Sánchez, Wear behavior in Al 2024–CNTs composites synthesized by mechanical alloying, Wear 292–293 (2012) 169–175.