- Email: [email protected]
Wear of thermal spray deposited low carbon steel coatings on aluminum alloys
Wear 251 (2001) 1023–1033
Wear of thermal spray deposited low carbon steel coatings on aluminum alloys A. Edrisy a , T. Perry b , Y.T. Cheng b , A.T. Alpas a,∗ a
Mechanical, Automotive and Materials Engineering, University of Windsor, Windsor, Ont., Canada N9B 3P4 b General Motors Research and Development Center, 30500 Mound Road, Warren, MI 48090-9055, USA
Abstract Sliding wear behavior of low carbon steel coatings deposited on 319 Al alloy substrates using a plasma transfer wire arc (PTWA) thermal spraying process was studied. The coatings had a layered microstructure consisting of steel splats and FeO veins (0.5–3.0 m thick) between them. Wear tests were performed using a pin-on-disc type wear tester within a load range of 10–75 N and a sliding speed range of 0.2–2.5 m/s against tool steel pins in a dry air atmosphere (7–10% RH). The wear rates, frictional forces, and surface temperatures were measured as a function of the applied load and sliding speed. In constant load tests, the wear rates decreased with increasing sliding velocity. The worn surfaces and the wear debris were characterized with optical microscopy (OM), scanning electron microscopy (SEM), energy dispersive spectroscopy (EDS), X-ray diffraction (XRD), and microhardness. It was found that the wear rates and mechanisms could be divided between four loading and velocity conditions. At low load and velocity, oxidation (formation of Fe2 O3 ) was the main wear mechanism. The highest wear rates were associated with severe deformation of the steel splat tips and eventually splat fracture and fragmentation and also formation of a mixture of iron oxides that occurred at low velocities and high loads. At high loads and velocities the wear rates decreased, this was associated with the formation of thick protective oxide film and hardening of the sliding surface. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Plasma transfer wire arc; Steel coatings; Thermal spray deposition; Sliding wear; Wear mechanisms
1. Introduction Replacing cast iron parts by ones made of cast aluminum alloys provides a possible cost-effective method of reducing the mass of many automotive components. However, poor wear resistance and low seizure load of unprotected 319 or 356 Al–Si based alloys [1] are a major drawback for applications that require sliding contact. The wear resistance of cast aluminum parts can be improved by protecting the sliding surfaces with liners (e.g. cast iron, metal matrix composite, etc.) or coatings. There are many technologies available to provide hard coatings for aluminum, in particular, thermal spray technologies are of significant interest for the automotive industry. Thermal spray coatings can be deposited with relatively short cycle times, and the heat input and resulting distortion are lower than many CVD processes. Compared to cast iron liners, ferrous thermal spray coatings on cast aluminum components have some advantages, such as lighter weight, possible reduced cost, and improved scuffing resistance [2]. ∗ Corresponding author. Tel.: +1-519-253-4232/ext. 2602; fax: +1-519-973-7007. E-mail address: [email protected] (A.T. Alpas).
Thermal spray metallic coatings have a lamellar microstructure consisting of splats, resulting from flattening of molten metal droplets as they hit the surface. They may also contain oxide layers and inclusions between the splats. The oxide inclusions form as a result of oxidation of molten metal droplets while in-flight. The oxide layers between the splats are probably the result of both surface oxidation of the molten metal droplets in-flight, and continued oxidation of the splats on the surface as they solidify. As the splats build up, they may also trap gas in the valleys of the roughened surfaces between adjacent splats, thereby forming micro pores upon solidification [3–5]. It has been reported that the sliding wear of thermal sprayed steel coatings could be attributed to splat delamination [6] due to the ‘weak links’ caused by the oxide veins [7]. The high wear rates were generally associated with the formation and propagation of subsurface cracks within the oxide veins, resulting in removal of whole splats during the sliding process. With the exceptions of splat delamination and mechanisms related to the laminated microstructures of these coatings, there are some similarities between the wear behavior of thermal spray low carbon steel coatings and low carbon steels. According to Welsh [8], when pins of plain carbon steel (0.12 to 0.78% C) were rubbed
0043-1648/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 3 - 1 6 4 8 ( 0 1 ) 0 0 7 1 8 - 9
1024
A. Edrisy et al. / Wear 251 (2001) 1023–1033
on rings of the same material, at a critical sliding speed, which depended on the composition of steel tested, the high wear rates (severe wear) suddenly diminished to low values (mild wear). This was attributed to a self-induced quench hardening process, as a result of frictional heating of asperities [8,9]. Archard [10], by studying the wear behavior of a 0.52% C steel in detail, found critical speeds for the severe to mild wear transitions at two loads. Using calculations he showed that at those critical speeds the surface temperature was high enough for a ferrite–austenite transformation and subsequent quench hardening. More recently, Lim and Ashby [11] studied the wear of steels using a wear map approach, they classified the wear mechanisms by accounting for the frictional heating and calculating the flash temperatures for four mechanisms. As will be shown here, among these wear mechanisms observed, the oxidation-dominated wear and severe plastic deformation induced wear appear to be most closely related to the wear of thermal spray low carbon steel coatings. A detailed review of oxidational wear was given by Quinn [12]. The work reported in this paper is a part of a larger project whose goal is to select the appropriate material and spray parameters that will provide coatings that have the necessary wear properties to replace the cylinder liners used in aluminum engines. The paper focuses on the dry sliding wear behavior of low carbon steel coatings produced using a plasma transferred wire arc (PTWA) process. 2. Materials and experimental procedures
were 250 ± 10 m thick. Deposition parameters are listed in Table 1. The surface roughness (Ra ) of these coatings was measured using a stylus type surface profilometer. The average roughness of 10 different samples was 18 ± 2 m. The structure and chemistry of a typical coating is shown in Fig. 1. For metallographic preparation of the cross-sections, a fast curing epoxy resin mixed with hardener was used as a mounting media. To avoid oxidation of the steel coatings anhydrous ethanol was used as lubricant after the 400-grit stage of grinding. Samples were then polished on polishing cloths impregnated with 3, 1 and 0.25 m diamond paste, respectively. Fig. 1(a) is a back-scattered SEM photograph that shows steel splats, 0.5–3.0 m thick iron oxide veins between the steel splats, and several micropores. The energy dispersive spectroscopy (EDS) analysis (Fig. 1(b) and (c)) of the different phases seen in Fig. 1(a) along with the XRD results in Fig. 1(d) confirmed that microstructure consisted of elemental Fe and only FeO type iron oxide. The microhardness of the PTWA coatings was calculated by averaging 10 Vickers microhardness indentations made on the top surface of polished coatings with a mean roughness of 0.34±0.1 m using a normal load of 25 g. The coating hardness determined in this way was 310 ± 10 kg/mm2 . The mean indentation depth at this load was 5 m, which was a small fraction of the total coating thickness. The mean Vickers indentation hardness of the 319 Al substrate was measured on uncoated polished surface using a 500 g load and was 90 ± 10 kg/mm2 . Some properties of the PTWA low carbon steel coatings produced for this work are summarized in the Table 2.
2.1. Microstructure, composition and properties of coatings
2.2. Wear tests
The stock for the 319 Al substrates was cast in the GM Research and Development Foundry, Warren, MI as 25.4 mm thick chill blocks and cut in the form of square coupons with dimension 25 mm × 25 mm × 5 mm, using a wire electric discharge machine (EDM). The surfaces were roughened to an average surface roughness of R a = 10 ± 2 m with an 80 grit sand blaster operating at a pressure of 0.62 MPa prior to deposition to provide mechanical interlocking for the coating to adhere to the substrate. The coatings were prepared using a commercial PTWA gun. The wire feed stock for the PTWA gun was mild steel with a nominal 1020 composition. The gun was operated at a nozzle to substrate distance of 49 mm, The plasma jet included molten iron droplets of 100–300 m diameter, traveling at a nominal speed of 200 m/s. The coating was built up by translating the gun over the coupon surface multiple times to produce coatings that
To eliminate the initial high wear rate normally caused by the coating roughness, and to ensure that the sample surfaces were flat, an automatic polishing machine was used to polish the wear test sample surfaces. With the exception of some small areas, where the coatings contained large amounts of porosity, the average surface roughness was decreased from 18 ± 2 to 0.38 ± 0.08 m. The wear tests were performed using a pin-on-disc sliding wear apparatus. The apparatus comprised a variable speed rotating shaft arrangement to which a stainless steel sample holder was attached. A vertical loading arm, which was attached to the pin counterface, was lowered on to the rotating coated samples to produce a circular wear track of 16 mm average diameter. The pin contact geometry used was a flat-on-flat configuration. An enclosure around the machine equipped with a humidifier and dehumidifier provided the opportunity of running tests
Table 1 PTWA deposition parameters Wire composition
Wire diameter (mm)
Wire feed rate (m/h)
Plasma gas
Spray distance (mm)
Carrier gas
SAE 1020 steel (0.2% C)
1.5
4.5
65% Ar; 35% H2
49
Air
A. Edrisy et al. / Wear 251 (2001) 1023–1033
1025
Fig. 1. Structure and chemistry of an unworn PTWA iron coating. (a) The back-scattered SEM image of a polished cross-section of the coating (unetched). Three distinct regions are evident in this photograph: the light areas, labelled (b) in the image are iron, as is shown in the EDS spectrum of (b); the dark grey regions labelled (c) are Fe and O, as is shown in the EDS spectrum of (c); the round black regions are porosity (d). (d) is the XRD spectrum of this sample showing the presence of iron and (FeO).
under controlled atmospheric conditions. The data presented here were taken with an air atmosphere with humidity of 7–10%. The counterface pin material was an AISI type M2 high-speed tool steel with the following composition in weight percent: 0.8 C, 4.0 Cr, 5.0 Mo, 6.0 W, 2.0 V and the balance Fe. The hardness of the AISI M2, measured using a load of 500 g was of 910 ± 10 kg/mm2 . The diameter and the length of the pin were 5.0 and 30 mm, respectively. Each of the experiments was performed at set sliding speeds between 0.2 and 2.5 m/s and at constant loads
between 10 and 75 N. The wear tests were run to a constant sliding distance of 5000 m. Prior to wear testing, both the pin and the coated samples were ultrasonically cleaned in acetone, left for 24 h under vacuum and weighed to ±0.0001 g using an electronic balance. After each test the specimens were cleaned of loose debris using compressed air and weighed to determine the amount of mass change during the test. Wear rates of both the PTWA coatings and the pins were obtained by dividing mass loss by the total sliding distance (5000 m). A 0.5 m diameter K-type ungrounded thermocouple probe was placed in a 0.6 m
Table 2 Properties of the PTWA low carbon steel coating Thickness
Hardness (HV25 g )
Roughness (after deposition process)
Roughness (before wear test)
250 ± 10 m
310 ±
18 ± 2 m
0.38 ± 0.08 m
10 kg/mm2
1026
A. Edrisy et al. / Wear 251 (2001) 1023–1033
wide vertical slot, cut by a diamond saw, in the end of each pin. The tip of the probe rested in the slot approximately 100–200 m above the pin-disc contact region and was used to monitor changes in bulk surface temperature (Tb ) generated by frictional heating at the sliding interface. The friction measurements were also recorded via a software interface and a load cell/torque arm arrangement, which was attached to the drive motor. A scanning electron microscope (JEOL 5800 LV SEM) equipped with EDS, and an XRD (Siemens D-500 with Cu K␣ tube) were used to characterize the composition (Rietveld refinement) and morphology of the worn surfaces, the cross-sections, and the wear debris particles.
3. Results 3.1. Wear rates The variation of wear rates of the coatings with sliding velocity at loads of 5, 10, 25, 50 and 75 N is shown in Fig. 2. The wear rates were high at low sliding velocity but decreased with increasing sliding velocity. This effect became more pronounced at high loads, especially at 50 and 75 N. At 50 N load, for the tests performed at a low sliding velocity of 0.2 m/s, the measured wear rate was 30.52×10−6 g/m; by increasing the sliding velocity to 2.5 m/s the wear rate decreased significantly to 7.52 × 10−6 g/m. This decrease in the wear rate with velocity indicates that in this velocity range the PTWA low carbon steel coating was subjected to a wear transition from a severe form of wear at low speeds to a mild form of wear at speeds approximately above 1.0 m/s.
Fig. 2. The variation of wear rates plotted against sliding velocity for four different normal loads. The trend is for decreased wear rates at increased velocity.
3.2. SEM observations of worn surfaces Fig. 3 is the back-scattered SEM micrograph of the worn surface of a sample tested at the highest load (75 N) for 0.5 m/s sliding velocity (low velocity). About 20–30% of the wear tracks were covered by the oxide rich films, which are the dark gray regions shown in Fig. 3. The EDS spectra of the dark gray regions in Fig. 3 confirmed that they included iron and oxygen. Attempts were made to obtain the XRD spectra directly from selected areas on the wear
Fig. 3. The back-scattered SEM micrograph of a sample worn at 75 N load and 0.5 m/s. The dark grey regions are thin oxide rich films, which are smeared over the surface of the steel splats (medium grey regions). The micrograph also shows crack formation along the ridges of steel splats.
A. Edrisy et al. / Wear 251 (2001) 1023–1033
1027
Fig. 4. The back-scattered SEM micrograph of a sample worn at 50 N load and 2.5 m/s. The micrograph in Fig. 4 is taken from a region, which shows evidence for deformation mostly in the form of surface grooving of both the oxide rich layers and the iron splats (note the micrograph in Fig. 4 was deliberately selected to illustrate the deformation pattern and the area fraction of oxide rich layers is less then the average coverage).
tracks in this and other samples, but the surface layers were too thin and the coverage was incomplete to resolve reliable information about the state of oxidation of these layers. However, the XRD spectra of the debris particles detached from the worn surfaces provided information about the oxides formed during sliding wear, and these are presented in Section 3.3. The top surfaces of steel splats (medium gray regions) were flattened as a result of plastic deformation during wear and extruded in the sliding direction. An important aspect of wear at high loads and low velocities is the fracture of the edges of the highly deformed steel splats. Metallographic evidence for the fracture of the edges of the steel splats after wear is shown in Fig. 3. Note that the highest wear rates were obtained under these conditions (high loads and low velocities), where the oxide rich films were thin and the edges of most steel splats were fractured. The worn surfaces of the samples tested at high loads and high velocities were significantly different. The primary difference was in the area fraction of iron oxide rich layers on the worn surfaces. About 70% of the wear tracks of the high load and high velocity samples were covered by relatively thick oxide rich layers, whose average thicknesses varied between 1 and 3 m. In contrast, the oxide rich films on the high load and low velocity samples were thinner (<1 m) and discontinuous. The micrograph in Fig. 4 is taken from a sample tested at 50 N and 2.5 m/s and shows evidence for deformation, mostly in the form of surface grooving of both the oxide rich layers and the iron splats (note that the micrograph in Fig. 4 was deliberately selected to illustrate the deformation pattern and the area fraction of oxide rich layers is less than the average coverage). The steel splats on the other hand, although severely deformed appeared to be much less susceptible to fracture.
Approximately 50% of the worn surfaces of the samples tested at low loads and high velocities (e.g. tested at 10 N and 2.5 m/s) were covered by the iron oxide rich layers as shown in Fig. 5 and these layers typically extended over the top of the steel splats. SEM evidence for steel splat fracture was very rare in this case. Worn surfaces of the samples tested at the same load (10 N) but at low velocities were similar to that shown in Fig. 5, but the surface areas covered by the oxide rich layers were less. In the samples tested at 10 N and 0.5 m/s, about 20% of the worn surface were covered by the oxide rich layers. At low loads the oxide rich layers were less compact and had a porous appearance. The cross-sections of samples occasionally revealed that steel splats adjacent to the worn surface delaminated along the FeO veins within the coating causing removal of entire individual splats as a whole. However, this was not a common observation. 3.3. XRD spectra of wear debris The XRD spectra of the wear debris from samples tested at three different test conditions are shown in Fig. 6(a) (10 N and 0.5 m/s); (b) (50 N and 0.5 m/s); and (c) (50 N, 2 m/s). Although some oxide peaks overlap, each oxide has unique peaks that do not overlap with the others, making unequivocal identification of these particular oxides possible in the debris. The quantitative phase analysis results obtained from Rietveld refinement of the spectra of Fig. 6 are summarized in Table 3. These samples spanned the different ranges of wear rates (Fig. 2) and were expected to exhibit different wear mechanisms. The wear rate was too low to collect a sufficient amount of debris at low load and high velocity. The XRD spectrum of the debris at 10 N load and 0.5 m/s
1028
A. Edrisy et al. / Wear 251 (2001) 1023–1033
Fig. 5. The back-scattered SEM micrograph of a sample worn at 10 N load and 2.5 m/s. The dark grey extended regions are an iron oxide film that is smeared over the surface of the steel splats. There is relatively little evidence of cracking or fracture damage of the steel splats.
velocity, representing low load and low velocity conditions, shows the peaks for hematite (Fe2 O3 ) and few small peaks of ferrite (Fig. 6(a)). The quantitative phase analysis shown in Table 3 confirms that at low loading and velocity conditions Fe2 O3 was the predominant constituent (>99%) of the wear debris. As shown in Table 3, by increasing the load to 50 N, at the same velocity (0.5 m/s), the proportion of Fe2 O3 in the debris decreased. The XRD spectrum in Fig. 6(b) shows that at high load and low velocity conditions the debris consisted of a mixture of three types of iron oxides: Fe2 O3 , Fe3 O4 , and FeO. It is important to emphasize that there was also a significant amount of metallic ferrite in the debris. Referring to Fig. 6(c) and Table 3, by increasing the sliding velocity from 0.5 to 2 m/s at a high load (50 N) the constituents of the debris remained the same, consisting of ferrite and three types of iron oxides, namely
Table 3 Quantitative XRD phase analysis of the debris (weight fraction) at three different loading conditionsa Phase
10 N, 0.5 m/s
50 N, 0.5 m/s
50 N, 2 m/s
Ferrite (Fe) Hematite (Fe2 O3 ) Magnetite (Fe3 O4 ) Wuestite (FeO)
0.02 99.97 0.01 0.00
15.38 41.43 20.30 22.89
6.70 60.31 22.13 10.85
a For all the spectra analyzed the estimated S.D. was always better than 2 wt.% in some case it was better than 0.5%.
Fe2 O3 , Fe3 O4 , and FeO but the relative amounts of the oxide phases changed slightly. It was also noted that the percentage of ferrite in the debris was considerably smaller. 3.4. Hardness of worn surfaces The variations of the average microhardness of the coating after wear testing with sliding velocity at the constant loads of 25, 50 and 75 N are shown in Fig. 7. For samples where there was an oxide layer on the wear track, the hardness data were obtained from regions where the iron was exposed. The hardness of the wear track was always higher than the unworn area of the coatings (310 kg/mm2 ) under all conditions. For example, the average worn surface hardness of samples tested at 25 N and 0.2 m/s was about 400 kg/mm2 . The increase of hardness after the wear test could be attributed to the work hardening of the sliding surfaces as a result of severe plastic deformation of the steel splats at or near the contact surface. The effect of surface hardening was more pronounced at higher loads. For example a surface hardness of 550 kg/mm2 was measured at 75 N (0.5 m/s). In fact, the worn surface hardness increased as the test conditions became more severe. For the tests conducted at high loads and velocities the hardness of wear tracks increased to exceptionally high values of over 700 kg/mm2 . At 75 N and 2.5 m/s, for example, the average hardness of the steel splats on the contact surface reached 800 kg/mm2 . These very signifi-
A. Edrisy et al. / Wear 251 (2001) 1023–1033
1029
Fig. 6. The XRD spectra of the wear debris collected from the following conditions: (a) low load, low velocity (10 N, 0.5 m/s); (b) high load, low velocity (50 N, 0.5 m/s); (c) high load, high velocity (50 N, 2 m/s). The oxide labels in brackets next to an X-ray peak imply that this peak belong to more than one type of oxide.
cant hardening effects cannot be explained solely by work hardening of the steel splats and will be considered in Section 4. 3.5. Temperature and coefficient of friction measurements Fig. 8(a) shows the variation of sliding induced temperature rises and coefficients of friction with sliding velocity for samples tested at a constant load of 50 N. Fig. 8(b) shows the variation of sliding induced temperature rises and coefficients of friction with normal load for samples tested at constant velocity of 2.5 m/s. In both cases the coefficient of friction decreased as the load and the speed was increased.
As shown in Figs. 8(a) and (b) the surface temperature increased significantly with the sliding velocity (to 250◦ C at 2.5 m/s) and the load (to 340◦ C at 75 N), presumably increasing the rate of surface oxidation. This was consistent with the SEM observations that the iron oxide rich layers on the wear tracks were thicker and more continuous when the test loads and speeds were high, hence, reducing the coefficient of friction. It should be noted that the temperatures reported in Fig. 8 were the average temperatures of a region 100–200 m above the contact surfaces. It is expected that the contact temperatures at the asperity tips (flash temperatures) could readily reach values about three times higher than these measured values [13].
1030
A. Edrisy et al. / Wear 251 (2001) 1023–1033
The salient features of the experimental observations are summarized in Table 4. The wear behavior of the coatings for each of the four conditions indicated in this table will be discussed in Section 4. 4. Discussion
Fig. 7. The variation of hardness with sliding velocity for high loads. The hardness measurements were taken from regions of the wear track where the steel coating was exposed through any oxide deposit.
As shown in Figs. 2, the wear rates of the coatings at constant load decreased with increasing sliding velocity up to 1.0–1.5 m/s and above that the wear rates were almost constant. This behavior became more pronounced at high loads, such as at 50 and 75 N, where the wear rates were higher initially and decreased faster than those at lower loads. From Fig. 2 four wear regimes can be identified amongst the testing conditions. These are defined as low velocity, ≤1 m/s, high velocity, ≥1 m/s, and low load, ≤20 N, and high load, ≥20 N. These regimes were defined according to the differences in the wear rates as well as in the chemical compositions and/or microstructures of the worn surfaces and the debris and will be considered separately below. 4.1. High loads and low velocities
Fig. 8. (a) The variation of sliding induced temperature rise of the pin and the coefficient of friction with sliding velocity for samples tested at constant load of 50 N. (b) The variation of the friction induced temperature rise of the pin and the coefficient of friction with normal load for samples tested at constant velocity of 2.5 m/s.
The highest coating wear rates were measured in tests performed under the conditions of high loads, and low velocities (Fig. 2). The wear debris collected under these conditions was magnetic, with a dark brown color. The XRD spectrum of the wear debris of a sample tested at 50 N load and 0.5 m/s showed the presence of ferrite, Fe2 O3 , Fe3 O4 , and FeO (Table 3). Although the peak temperature rise of the pin was approximately 75◦ C Fig. 8(a), the presence of Fe3 O4 , in the wear debris suggests that the temperature at the asperity contact area (flash temperature) must have been significantly higher (since, Fe3 O4 grows under static conditions at above over 450◦ C [14,15]). The back-scattered SEM image in Fig. 3 shows the morphology of the worn surface of sample tested at 0.5 m/s sliding velocity and 75 N where a thin layer of iron oxide formed across the tops of steel splats. More significantly, Fig. 3 also shows evidence for fracture of steel splats. The wear mechanism at high loads and low velocities must account for the presence of the iron particles in the wear debris. The iron fragments were the results of fracture of splats due to severe plastic deformation at the tips of the splats. An interpretation of fracture is shown schematically in Fig. 9. At the first stages of sliding (Fig. 9(a)), steel splats adjacent to contact surfaces deform and elongate in the direction of sliding. At subsequent stages more deformation is induced, especially at the elongated tips of splats (b). Exhaustion of ductility of the material at the splat tips causes fracture and fragmentation as shown in (c). Examples of fractured splat tips can be seen in Fig. 3. Occasionally, repeated fracture and fragmentation of splats progresses to a degree where the entire splat is lost. Porosity or other roughness in the surface makes this mechanism worse. Steel splats adjacent to surface depressions can plastically deform and elongate
A. Edrisy et al. / Wear 251 (2001) 1023–1033
1031
1032
A. Edrisy et al. / Wear 251 (2001) 1023–1033
either OM or SEM. The XRD spectrum of the wear debris of the sample tested at 10 N load and 0.5 m/s sliding velocity confirmed the formation of Fe2 O3 (Fig. 6(a)) the quantitative phase analysis showed that the major constituent of the debris was Fe2 O3 with a negligible amount of ferrite. (Table 3). Under static conditions Fe2 O3 is the equilibrium phase that forms under ambient conditions [16]. This is consistent with the measured bulk temperature of 60◦ C (Fig. 8) and the wear debris collected was a non-magnetic powder with a dark orange color indicative of hydrated Fe2 O3 . The wear rates were lower than those encountered at the same low velocity levels but at higher loads. 4.3. High velocities and low loads
Fig. 9. A schematic diagram illustrating a mechanism for deformation and fracture of steel splats: (a) shows a cross-section of the unworn surface; (b) shows a cross-section of plastically deformed splats elongated along the sliding direction; (c) is a surface view showing fracture at the edges of the splats.
At high velocity and low loads the wear rate had the lowest value compared to the other regimes (Fig. 2). The SEM micrograph (Fig. 5) of the worn surface of a sample tested under these conditions shows a thin deposit of oxide. However, coverage was more significant than at low load and low velocity conditions and extended over the top of the steel splats. The measured pin temperature for the sample tested at 10 N and 2.5 m/s was about 110◦ C, which was higher than the sample tested at low velocity and low load. Because, the wear rate was very low, insufficient wear debris could be collected for XRD investigation. It can be suggested that several favorable factors, including the formation of almost continuous oxide rich films, hardening of the coating during wear, and lack of any significant iron splat fracture and delamination, are among the possible reasons why the wear rates are the lowest under these conditions. 4.4. High velocities and high loads
into these regions much more easily than in dense smooth regions that are self-supporting. Another wear mechanism related to the lamellar structure of the coatings is the splat delamination. Oxide layers between splats are the weak links in many spray deposited coatings [7]. In [6], the high wear rates were generally associated with the formation and propagation of subsurface cracks within the oxide in the thermal spray deposited coatings, resulting in removal of complete steel splats during the sliding process. The splat orientation and surface waviness were suggested to have an influence on the delamination mechanism. Delamination was shown to be easier when the splats were parallel to the coating surface and more difficult when they were wavy and not parallel to interface [6]. Splat delamination was occasionally observed under the present experimental conditions but was not a dominant wear mechanism for the PTWA low carbon steel coatings. 4.2. Low loads and low velocities At low load and velocity oxidation appeared to be the main wear mechanism. Splat fracture was rarely observed in
At high velocities and high loads the wear rates are lower than at low velocities (Fig. 2). In fact the wear rates were about 50% lower than those measured at similar loads but using low test speeds, which sounds counter intuitive. However, the experimental data suggested that high surface temperatures played an important role to keep the wear rates low under these most severe testing conditions. The high surface temperatures may have reduced wear rates in two ways, (i) by producing thick oxide rich layers (Fig. 4) that reduce the coefficient of friction and probably suppressing the splat tip fracture mechanism, (ii) by increasing the hardness of the steel splat surfaces during wear (Fig. 7). The pin temperature (Fig. 8) shows that the sliding induced temperature rise was influenced more by sliding velocity than by load. The maximum bulk temperature rise occurred for the sample tested at 75 N and 2.5 m/s, where the increase was about 390◦ C. The wear track of the high velocity and high load sample had regions (about 70%) that were covered with a thick oxide rich layer whose thickness reached between 1 and 3 m at some points. Fig. 8 shows that by increasing velocity from 0.5 to 2.5 m/s at 50 N load the coefficient
A. Edrisy et al. / Wear 251 (2001) 1023–1033
of friction decreased from approximately 0.8 to 0. 6 this is also consistent with the formation of thick oxide rich film. The XRD spectrum of the wear debris of the sample tested at 50 N load and 2 m/s in Fig. 6(c) showed the presence of all three oxides. Although these phases were the same as those of the debris from low velocity and high load tests, the quantitative phase analysis shows the relative amount of ferrite at high velocity was lower. Evidence from SEM and OM suggested that the splat tip fracture mechanism that was responsible for the high wear rates at low load and velocity conditions was not as significant when the sliding speed was high. The hardness of some parts of the wear tracks in samples tested at high loads, e.g. 50 and 75 N and 2.5 m/s sliding velocity increased to over 800 kg/mm2 , from the initial 310 kg/mm2 for the unworn surface. Welsh attributed the excessive hardening of plain carbon steels during dry sliding to a martensitic phase transformation induced by temperature fluctuations on the sliding surfaces. It was also mentioned that when the temperature was high enough to allow an austenite transformation, the probability of absorption of nitrogen or the other constituents of the atmosphere was high [8,9]. The origin of this hardening effect in the PTWA low carbon steel coatings was not revealed by the wear experiments. The temperature should exceed 860◦ C for any martensite to form in 1020 steel by a conventional heating and quenching mechanism. In an attempt to simulate the effect of flash temperature at the asperities, PTWA coatings were instantaneously heated by electrical resistance (spark) heating and self-quenched in a bell jar under argon atmosphere. The resulting microhardness values were similar to those generated by sliding wear (750 kg/ mm2 ). The effect of surface hardening during wear should be studied in more detail. 5. Conclusions The wear rates of thermal spray deposited low carbon steel coatings produced by a PTWA process were measured at different load and velocity conditions. The wear mechanisms of the coating were associated with surface oxidation, splat tip fracture and surface hardening during wear. Splat delamination or removal of individual steel splats as a result of crack growth along the FeO veins within the coating was seldom observed. The wear rates and mechanisms can be classified as follows: 1. At low load and velocity, oxidation (formation of Fe2 O3 ) was the main wear mechanism.
1033
2. At low velocity and high load the high wear rate was associated with severe deformation of the steel splat tips and eventually splat fracture and fragmentation. In addition, tribological layers consisting of a mixture of Fe2 O3 , FeO, and Fe3 O4 formed on the contact surfaces. 3. At high load and velocity the wear rates decreased. Evidence was found for two different mechanisms that could account for the relatively low wear rates. The first mechanism was the formation of a thick protective mixed oxide layer on the wear track. The second mechanism was a pronounced hardening of the iron coating. 4. The wear rates were the lowest at high velocity and low load conditions, where there was no evidence for the splat tip fracture mechanism and the surfaces were covered with oxide rich tribolayers.
Acknowledgements The authors would like to thank Michael Balogh for his expertise with Rietveld analysis, T.W. Capehart for reading the manuscript, and John Robinson for his assistance with metallography and material characterization techniques. References [1] S. Wilson, A.T. Alpas, Wear 212 (1997) 41–49. [2] C.C. Berndt, S. Sampath (Eds.), Thermal Spray Industrial Applications, ASM International, Materials Park, OH, 1994. [3] L. Pawlowski, P. Fauchais, Int. Mater. Rev. 37 (6) (1992) 271–289. [4] V.V. Sobolev, J.M. Guilemany, J. Nutting, J.R. Miquel, Int. Mater. Rev. 42 (3) (1997) 117–136. [5] B.S. Schorr, K.J. Stein, A.R. Marder, Mater. Character. 42 (1999) 93–100. [6] S.E. Hartfield-Wunsch, S.C. Tung, in: C.C. Berndt, S. Sampath (Eds.), Thermal Spray Industrial Applications, ASM International, Materials Park, OH, 1994, pp. 19–32. [7] A. Rabiei, D.R. Mumm, J.W. Hutchinson, R. Schweinfest, M. Ruhle, A.G. Evans, Mater. Sci. Eng. A 269 (1999) 152–165. [8] N.C. Welsh, Royal Society of London, Philos. Trans. Series A 257 (1964) 34–52. [9] N.C. Welsh, J. Appl. Phys. 28 (9) (1957) 960–968. [10] J.F. Archard, Wear 2 (1952) 438. [11] S.C. Lim, M.F. Ashby, Acta Metallurgica. 35 (1987) 1–24. [12] T.F.J. Quinn, Tribol. Int. 16 (5) (1983) 257–271. [13] M.F. Ashby, J. Abulawi, H.S. Kong, Tribol. Trans. 34 (1991) 577–587. [14] I.M. Hutchings, Tribology, Friction and Wear of Engineering Materials, CRC Press, Cambridge, 1992, pp. 103–106. [15] T.F.J. Quinn, Oxidational wear in friction, lubrication, and wear technology, ASM Handbook, Vol. 18, ASM International, Metals Park, OH, 1998, pp. 280–288. [16] A. Muan, E.F. Osborn, Phase Equilibria Among Oxides in Steelmaking, Addison-Wesley, Reading, MA, 1965, p. 27.
Wear of thermal spray deposited low carbon steel coatings on aluminum alloys A. Edrisy a , T. Perry b , Y.T. Cheng b , A.T. Alpas a,∗ a
Mechanical, Automotive and Materials Engineering, University of Windsor, Windsor, Ont., Canada N9B 3P4 b General Motors Research and Development Center, 30500 Mound Road, Warren, MI 48090-9055, USA
Abstract Sliding wear behavior of low carbon steel coatings deposited on 319 Al alloy substrates using a plasma transfer wire arc (PTWA) thermal spraying process was studied. The coatings had a layered microstructure consisting of steel splats and FeO veins (0.5–3.0 m thick) between them. Wear tests were performed using a pin-on-disc type wear tester within a load range of 10–75 N and a sliding speed range of 0.2–2.5 m/s against tool steel pins in a dry air atmosphere (7–10% RH). The wear rates, frictional forces, and surface temperatures were measured as a function of the applied load and sliding speed. In constant load tests, the wear rates decreased with increasing sliding velocity. The worn surfaces and the wear debris were characterized with optical microscopy (OM), scanning electron microscopy (SEM), energy dispersive spectroscopy (EDS), X-ray diffraction (XRD), and microhardness. It was found that the wear rates and mechanisms could be divided between four loading and velocity conditions. At low load and velocity, oxidation (formation of Fe2 O3 ) was the main wear mechanism. The highest wear rates were associated with severe deformation of the steel splat tips and eventually splat fracture and fragmentation and also formation of a mixture of iron oxides that occurred at low velocities and high loads. At high loads and velocities the wear rates decreased, this was associated with the formation of thick protective oxide film and hardening of the sliding surface. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Plasma transfer wire arc; Steel coatings; Thermal spray deposition; Sliding wear; Wear mechanisms
1. Introduction Replacing cast iron parts by ones made of cast aluminum alloys provides a possible cost-effective method of reducing the mass of many automotive components. However, poor wear resistance and low seizure load of unprotected 319 or 356 Al–Si based alloys [1] are a major drawback for applications that require sliding contact. The wear resistance of cast aluminum parts can be improved by protecting the sliding surfaces with liners (e.g. cast iron, metal matrix composite, etc.) or coatings. There are many technologies available to provide hard coatings for aluminum, in particular, thermal spray technologies are of significant interest for the automotive industry. Thermal spray coatings can be deposited with relatively short cycle times, and the heat input and resulting distortion are lower than many CVD processes. Compared to cast iron liners, ferrous thermal spray coatings on cast aluminum components have some advantages, such as lighter weight, possible reduced cost, and improved scuffing resistance [2]. ∗ Corresponding author. Tel.: +1-519-253-4232/ext. 2602; fax: +1-519-973-7007. E-mail address: [email protected] (A.T. Alpas).
Thermal spray metallic coatings have a lamellar microstructure consisting of splats, resulting from flattening of molten metal droplets as they hit the surface. They may also contain oxide layers and inclusions between the splats. The oxide inclusions form as a result of oxidation of molten metal droplets while in-flight. The oxide layers between the splats are probably the result of both surface oxidation of the molten metal droplets in-flight, and continued oxidation of the splats on the surface as they solidify. As the splats build up, they may also trap gas in the valleys of the roughened surfaces between adjacent splats, thereby forming micro pores upon solidification [3–5]. It has been reported that the sliding wear of thermal sprayed steel coatings could be attributed to splat delamination [6] due to the ‘weak links’ caused by the oxide veins [7]. The high wear rates were generally associated with the formation and propagation of subsurface cracks within the oxide veins, resulting in removal of whole splats during the sliding process. With the exceptions of splat delamination and mechanisms related to the laminated microstructures of these coatings, there are some similarities between the wear behavior of thermal spray low carbon steel coatings and low carbon steels. According to Welsh [8], when pins of plain carbon steel (0.12 to 0.78% C) were rubbed
0043-1648/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 3 - 1 6 4 8 ( 0 1 ) 0 0 7 1 8 - 9
1024
A. Edrisy et al. / Wear 251 (2001) 1023–1033
on rings of the same material, at a critical sliding speed, which depended on the composition of steel tested, the high wear rates (severe wear) suddenly diminished to low values (mild wear). This was attributed to a self-induced quench hardening process, as a result of frictional heating of asperities [8,9]. Archard [10], by studying the wear behavior of a 0.52% C steel in detail, found critical speeds for the severe to mild wear transitions at two loads. Using calculations he showed that at those critical speeds the surface temperature was high enough for a ferrite–austenite transformation and subsequent quench hardening. More recently, Lim and Ashby [11] studied the wear of steels using a wear map approach, they classified the wear mechanisms by accounting for the frictional heating and calculating the flash temperatures for four mechanisms. As will be shown here, among these wear mechanisms observed, the oxidation-dominated wear and severe plastic deformation induced wear appear to be most closely related to the wear of thermal spray low carbon steel coatings. A detailed review of oxidational wear was given by Quinn [12]. The work reported in this paper is a part of a larger project whose goal is to select the appropriate material and spray parameters that will provide coatings that have the necessary wear properties to replace the cylinder liners used in aluminum engines. The paper focuses on the dry sliding wear behavior of low carbon steel coatings produced using a plasma transferred wire arc (PTWA) process. 2. Materials and experimental procedures
were 250 ± 10 m thick. Deposition parameters are listed in Table 1. The surface roughness (Ra ) of these coatings was measured using a stylus type surface profilometer. The average roughness of 10 different samples was 18 ± 2 m. The structure and chemistry of a typical coating is shown in Fig. 1. For metallographic preparation of the cross-sections, a fast curing epoxy resin mixed with hardener was used as a mounting media. To avoid oxidation of the steel coatings anhydrous ethanol was used as lubricant after the 400-grit stage of grinding. Samples were then polished on polishing cloths impregnated with 3, 1 and 0.25 m diamond paste, respectively. Fig. 1(a) is a back-scattered SEM photograph that shows steel splats, 0.5–3.0 m thick iron oxide veins between the steel splats, and several micropores. The energy dispersive spectroscopy (EDS) analysis (Fig. 1(b) and (c)) of the different phases seen in Fig. 1(a) along with the XRD results in Fig. 1(d) confirmed that microstructure consisted of elemental Fe and only FeO type iron oxide. The microhardness of the PTWA coatings was calculated by averaging 10 Vickers microhardness indentations made on the top surface of polished coatings with a mean roughness of 0.34±0.1 m using a normal load of 25 g. The coating hardness determined in this way was 310 ± 10 kg/mm2 . The mean indentation depth at this load was 5 m, which was a small fraction of the total coating thickness. The mean Vickers indentation hardness of the 319 Al substrate was measured on uncoated polished surface using a 500 g load and was 90 ± 10 kg/mm2 . Some properties of the PTWA low carbon steel coatings produced for this work are summarized in the Table 2.
2.1. Microstructure, composition and properties of coatings
2.2. Wear tests
The stock for the 319 Al substrates was cast in the GM Research and Development Foundry, Warren, MI as 25.4 mm thick chill blocks and cut in the form of square coupons with dimension 25 mm × 25 mm × 5 mm, using a wire electric discharge machine (EDM). The surfaces were roughened to an average surface roughness of R a = 10 ± 2 m with an 80 grit sand blaster operating at a pressure of 0.62 MPa prior to deposition to provide mechanical interlocking for the coating to adhere to the substrate. The coatings were prepared using a commercial PTWA gun. The wire feed stock for the PTWA gun was mild steel with a nominal 1020 composition. The gun was operated at a nozzle to substrate distance of 49 mm, The plasma jet included molten iron droplets of 100–300 m diameter, traveling at a nominal speed of 200 m/s. The coating was built up by translating the gun over the coupon surface multiple times to produce coatings that
To eliminate the initial high wear rate normally caused by the coating roughness, and to ensure that the sample surfaces were flat, an automatic polishing machine was used to polish the wear test sample surfaces. With the exception of some small areas, where the coatings contained large amounts of porosity, the average surface roughness was decreased from 18 ± 2 to 0.38 ± 0.08 m. The wear tests were performed using a pin-on-disc sliding wear apparatus. The apparatus comprised a variable speed rotating shaft arrangement to which a stainless steel sample holder was attached. A vertical loading arm, which was attached to the pin counterface, was lowered on to the rotating coated samples to produce a circular wear track of 16 mm average diameter. The pin contact geometry used was a flat-on-flat configuration. An enclosure around the machine equipped with a humidifier and dehumidifier provided the opportunity of running tests
Table 1 PTWA deposition parameters Wire composition
Wire diameter (mm)
Wire feed rate (m/h)
Plasma gas
Spray distance (mm)
Carrier gas
SAE 1020 steel (0.2% C)
1.5
4.5
65% Ar; 35% H2
49
Air
A. Edrisy et al. / Wear 251 (2001) 1023–1033
1025
Fig. 1. Structure and chemistry of an unworn PTWA iron coating. (a) The back-scattered SEM image of a polished cross-section of the coating (unetched). Three distinct regions are evident in this photograph: the light areas, labelled (b) in the image are iron, as is shown in the EDS spectrum of (b); the dark grey regions labelled (c) are Fe and O, as is shown in the EDS spectrum of (c); the round black regions are porosity (d). (d) is the XRD spectrum of this sample showing the presence of iron and (FeO).
under controlled atmospheric conditions. The data presented here were taken with an air atmosphere with humidity of 7–10%. The counterface pin material was an AISI type M2 high-speed tool steel with the following composition in weight percent: 0.8 C, 4.0 Cr, 5.0 Mo, 6.0 W, 2.0 V and the balance Fe. The hardness of the AISI M2, measured using a load of 500 g was of 910 ± 10 kg/mm2 . The diameter and the length of the pin were 5.0 and 30 mm, respectively. Each of the experiments was performed at set sliding speeds between 0.2 and 2.5 m/s and at constant loads
between 10 and 75 N. The wear tests were run to a constant sliding distance of 5000 m. Prior to wear testing, both the pin and the coated samples were ultrasonically cleaned in acetone, left for 24 h under vacuum and weighed to ±0.0001 g using an electronic balance. After each test the specimens were cleaned of loose debris using compressed air and weighed to determine the amount of mass change during the test. Wear rates of both the PTWA coatings and the pins were obtained by dividing mass loss by the total sliding distance (5000 m). A 0.5 m diameter K-type ungrounded thermocouple probe was placed in a 0.6 m
Table 2 Properties of the PTWA low carbon steel coating Thickness
Hardness (HV25 g )
Roughness (after deposition process)
Roughness (before wear test)
250 ± 10 m
310 ±
18 ± 2 m
0.38 ± 0.08 m
10 kg/mm2
1026
A. Edrisy et al. / Wear 251 (2001) 1023–1033
wide vertical slot, cut by a diamond saw, in the end of each pin. The tip of the probe rested in the slot approximately 100–200 m above the pin-disc contact region and was used to monitor changes in bulk surface temperature (Tb ) generated by frictional heating at the sliding interface. The friction measurements were also recorded via a software interface and a load cell/torque arm arrangement, which was attached to the drive motor. A scanning electron microscope (JEOL 5800 LV SEM) equipped with EDS, and an XRD (Siemens D-500 with Cu K␣ tube) were used to characterize the composition (Rietveld refinement) and morphology of the worn surfaces, the cross-sections, and the wear debris particles.
3. Results 3.1. Wear rates The variation of wear rates of the coatings with sliding velocity at loads of 5, 10, 25, 50 and 75 N is shown in Fig. 2. The wear rates were high at low sliding velocity but decreased with increasing sliding velocity. This effect became more pronounced at high loads, especially at 50 and 75 N. At 50 N load, for the tests performed at a low sliding velocity of 0.2 m/s, the measured wear rate was 30.52×10−6 g/m; by increasing the sliding velocity to 2.5 m/s the wear rate decreased significantly to 7.52 × 10−6 g/m. This decrease in the wear rate with velocity indicates that in this velocity range the PTWA low carbon steel coating was subjected to a wear transition from a severe form of wear at low speeds to a mild form of wear at speeds approximately above 1.0 m/s.
Fig. 2. The variation of wear rates plotted against sliding velocity for four different normal loads. The trend is for decreased wear rates at increased velocity.
3.2. SEM observations of worn surfaces Fig. 3 is the back-scattered SEM micrograph of the worn surface of a sample tested at the highest load (75 N) for 0.5 m/s sliding velocity (low velocity). About 20–30% of the wear tracks were covered by the oxide rich films, which are the dark gray regions shown in Fig. 3. The EDS spectra of the dark gray regions in Fig. 3 confirmed that they included iron and oxygen. Attempts were made to obtain the XRD spectra directly from selected areas on the wear
Fig. 3. The back-scattered SEM micrograph of a sample worn at 75 N load and 0.5 m/s. The dark grey regions are thin oxide rich films, which are smeared over the surface of the steel splats (medium grey regions). The micrograph also shows crack formation along the ridges of steel splats.
A. Edrisy et al. / Wear 251 (2001) 1023–1033
1027
Fig. 4. The back-scattered SEM micrograph of a sample worn at 50 N load and 2.5 m/s. The micrograph in Fig. 4 is taken from a region, which shows evidence for deformation mostly in the form of surface grooving of both the oxide rich layers and the iron splats (note the micrograph in Fig. 4 was deliberately selected to illustrate the deformation pattern and the area fraction of oxide rich layers is less then the average coverage).
tracks in this and other samples, but the surface layers were too thin and the coverage was incomplete to resolve reliable information about the state of oxidation of these layers. However, the XRD spectra of the debris particles detached from the worn surfaces provided information about the oxides formed during sliding wear, and these are presented in Section 3.3. The top surfaces of steel splats (medium gray regions) were flattened as a result of plastic deformation during wear and extruded in the sliding direction. An important aspect of wear at high loads and low velocities is the fracture of the edges of the highly deformed steel splats. Metallographic evidence for the fracture of the edges of the steel splats after wear is shown in Fig. 3. Note that the highest wear rates were obtained under these conditions (high loads and low velocities), where the oxide rich films were thin and the edges of most steel splats were fractured. The worn surfaces of the samples tested at high loads and high velocities were significantly different. The primary difference was in the area fraction of iron oxide rich layers on the worn surfaces. About 70% of the wear tracks of the high load and high velocity samples were covered by relatively thick oxide rich layers, whose average thicknesses varied between 1 and 3 m. In contrast, the oxide rich films on the high load and low velocity samples were thinner (<1 m) and discontinuous. The micrograph in Fig. 4 is taken from a sample tested at 50 N and 2.5 m/s and shows evidence for deformation, mostly in the form of surface grooving of both the oxide rich layers and the iron splats (note that the micrograph in Fig. 4 was deliberately selected to illustrate the deformation pattern and the area fraction of oxide rich layers is less than the average coverage). The steel splats on the other hand, although severely deformed appeared to be much less susceptible to fracture.
Approximately 50% of the worn surfaces of the samples tested at low loads and high velocities (e.g. tested at 10 N and 2.5 m/s) were covered by the iron oxide rich layers as shown in Fig. 5 and these layers typically extended over the top of the steel splats. SEM evidence for steel splat fracture was very rare in this case. Worn surfaces of the samples tested at the same load (10 N) but at low velocities were similar to that shown in Fig. 5, but the surface areas covered by the oxide rich layers were less. In the samples tested at 10 N and 0.5 m/s, about 20% of the worn surface were covered by the oxide rich layers. At low loads the oxide rich layers were less compact and had a porous appearance. The cross-sections of samples occasionally revealed that steel splats adjacent to the worn surface delaminated along the FeO veins within the coating causing removal of entire individual splats as a whole. However, this was not a common observation. 3.3. XRD spectra of wear debris The XRD spectra of the wear debris from samples tested at three different test conditions are shown in Fig. 6(a) (10 N and 0.5 m/s); (b) (50 N and 0.5 m/s); and (c) (50 N, 2 m/s). Although some oxide peaks overlap, each oxide has unique peaks that do not overlap with the others, making unequivocal identification of these particular oxides possible in the debris. The quantitative phase analysis results obtained from Rietveld refinement of the spectra of Fig. 6 are summarized in Table 3. These samples spanned the different ranges of wear rates (Fig. 2) and were expected to exhibit different wear mechanisms. The wear rate was too low to collect a sufficient amount of debris at low load and high velocity. The XRD spectrum of the debris at 10 N load and 0.5 m/s
1028
A. Edrisy et al. / Wear 251 (2001) 1023–1033
Fig. 5. The back-scattered SEM micrograph of a sample worn at 10 N load and 2.5 m/s. The dark grey extended regions are an iron oxide film that is smeared over the surface of the steel splats. There is relatively little evidence of cracking or fracture damage of the steel splats.
velocity, representing low load and low velocity conditions, shows the peaks for hematite (Fe2 O3 ) and few small peaks of ferrite (Fig. 6(a)). The quantitative phase analysis shown in Table 3 confirms that at low loading and velocity conditions Fe2 O3 was the predominant constituent (>99%) of the wear debris. As shown in Table 3, by increasing the load to 50 N, at the same velocity (0.5 m/s), the proportion of Fe2 O3 in the debris decreased. The XRD spectrum in Fig. 6(b) shows that at high load and low velocity conditions the debris consisted of a mixture of three types of iron oxides: Fe2 O3 , Fe3 O4 , and FeO. It is important to emphasize that there was also a significant amount of metallic ferrite in the debris. Referring to Fig. 6(c) and Table 3, by increasing the sliding velocity from 0.5 to 2 m/s at a high load (50 N) the constituents of the debris remained the same, consisting of ferrite and three types of iron oxides, namely
Table 3 Quantitative XRD phase analysis of the debris (weight fraction) at three different loading conditionsa Phase
10 N, 0.5 m/s
50 N, 0.5 m/s
50 N, 2 m/s
Ferrite (Fe) Hematite (Fe2 O3 ) Magnetite (Fe3 O4 ) Wuestite (FeO)
0.02 99.97 0.01 0.00
15.38 41.43 20.30 22.89
6.70 60.31 22.13 10.85
a For all the spectra analyzed the estimated S.D. was always better than 2 wt.% in some case it was better than 0.5%.
Fe2 O3 , Fe3 O4 , and FeO but the relative amounts of the oxide phases changed slightly. It was also noted that the percentage of ferrite in the debris was considerably smaller. 3.4. Hardness of worn surfaces The variations of the average microhardness of the coating after wear testing with sliding velocity at the constant loads of 25, 50 and 75 N are shown in Fig. 7. For samples where there was an oxide layer on the wear track, the hardness data were obtained from regions where the iron was exposed. The hardness of the wear track was always higher than the unworn area of the coatings (310 kg/mm2 ) under all conditions. For example, the average worn surface hardness of samples tested at 25 N and 0.2 m/s was about 400 kg/mm2 . The increase of hardness after the wear test could be attributed to the work hardening of the sliding surfaces as a result of severe plastic deformation of the steel splats at or near the contact surface. The effect of surface hardening was more pronounced at higher loads. For example a surface hardness of 550 kg/mm2 was measured at 75 N (0.5 m/s). In fact, the worn surface hardness increased as the test conditions became more severe. For the tests conducted at high loads and velocities the hardness of wear tracks increased to exceptionally high values of over 700 kg/mm2 . At 75 N and 2.5 m/s, for example, the average hardness of the steel splats on the contact surface reached 800 kg/mm2 . These very signifi-
A. Edrisy et al. / Wear 251 (2001) 1023–1033
1029
Fig. 6. The XRD spectra of the wear debris collected from the following conditions: (a) low load, low velocity (10 N, 0.5 m/s); (b) high load, low velocity (50 N, 0.5 m/s); (c) high load, high velocity (50 N, 2 m/s). The oxide labels in brackets next to an X-ray peak imply that this peak belong to more than one type of oxide.
cant hardening effects cannot be explained solely by work hardening of the steel splats and will be considered in Section 4. 3.5. Temperature and coefficient of friction measurements Fig. 8(a) shows the variation of sliding induced temperature rises and coefficients of friction with sliding velocity for samples tested at a constant load of 50 N. Fig. 8(b) shows the variation of sliding induced temperature rises and coefficients of friction with normal load for samples tested at constant velocity of 2.5 m/s. In both cases the coefficient of friction decreased as the load and the speed was increased.
As shown in Figs. 8(a) and (b) the surface temperature increased significantly with the sliding velocity (to 250◦ C at 2.5 m/s) and the load (to 340◦ C at 75 N), presumably increasing the rate of surface oxidation. This was consistent with the SEM observations that the iron oxide rich layers on the wear tracks were thicker and more continuous when the test loads and speeds were high, hence, reducing the coefficient of friction. It should be noted that the temperatures reported in Fig. 8 were the average temperatures of a region 100–200 m above the contact surfaces. It is expected that the contact temperatures at the asperity tips (flash temperatures) could readily reach values about three times higher than these measured values [13].
1030
A. Edrisy et al. / Wear 251 (2001) 1023–1033
The salient features of the experimental observations are summarized in Table 4. The wear behavior of the coatings for each of the four conditions indicated in this table will be discussed in Section 4. 4. Discussion
Fig. 7. The variation of hardness with sliding velocity for high loads. The hardness measurements were taken from regions of the wear track where the steel coating was exposed through any oxide deposit.
As shown in Figs. 2, the wear rates of the coatings at constant load decreased with increasing sliding velocity up to 1.0–1.5 m/s and above that the wear rates were almost constant. This behavior became more pronounced at high loads, such as at 50 and 75 N, where the wear rates were higher initially and decreased faster than those at lower loads. From Fig. 2 four wear regimes can be identified amongst the testing conditions. These are defined as low velocity, ≤1 m/s, high velocity, ≥1 m/s, and low load, ≤20 N, and high load, ≥20 N. These regimes were defined according to the differences in the wear rates as well as in the chemical compositions and/or microstructures of the worn surfaces and the debris and will be considered separately below. 4.1. High loads and low velocities
Fig. 8. (a) The variation of sliding induced temperature rise of the pin and the coefficient of friction with sliding velocity for samples tested at constant load of 50 N. (b) The variation of the friction induced temperature rise of the pin and the coefficient of friction with normal load for samples tested at constant velocity of 2.5 m/s.
The highest coating wear rates were measured in tests performed under the conditions of high loads, and low velocities (Fig. 2). The wear debris collected under these conditions was magnetic, with a dark brown color. The XRD spectrum of the wear debris of a sample tested at 50 N load and 0.5 m/s showed the presence of ferrite, Fe2 O3 , Fe3 O4 , and FeO (Table 3). Although the peak temperature rise of the pin was approximately 75◦ C Fig. 8(a), the presence of Fe3 O4 , in the wear debris suggests that the temperature at the asperity contact area (flash temperature) must have been significantly higher (since, Fe3 O4 grows under static conditions at above over 450◦ C [14,15]). The back-scattered SEM image in Fig. 3 shows the morphology of the worn surface of sample tested at 0.5 m/s sliding velocity and 75 N where a thin layer of iron oxide formed across the tops of steel splats. More significantly, Fig. 3 also shows evidence for fracture of steel splats. The wear mechanism at high loads and low velocities must account for the presence of the iron particles in the wear debris. The iron fragments were the results of fracture of splats due to severe plastic deformation at the tips of the splats. An interpretation of fracture is shown schematically in Fig. 9. At the first stages of sliding (Fig. 9(a)), steel splats adjacent to contact surfaces deform and elongate in the direction of sliding. At subsequent stages more deformation is induced, especially at the elongated tips of splats (b). Exhaustion of ductility of the material at the splat tips causes fracture and fragmentation as shown in (c). Examples of fractured splat tips can be seen in Fig. 3. Occasionally, repeated fracture and fragmentation of splats progresses to a degree where the entire splat is lost. Porosity or other roughness in the surface makes this mechanism worse. Steel splats adjacent to surface depressions can plastically deform and elongate
A. Edrisy et al. / Wear 251 (2001) 1023–1033
1031
1032
A. Edrisy et al. / Wear 251 (2001) 1023–1033
either OM or SEM. The XRD spectrum of the wear debris of the sample tested at 10 N load and 0.5 m/s sliding velocity confirmed the formation of Fe2 O3 (Fig. 6(a)) the quantitative phase analysis showed that the major constituent of the debris was Fe2 O3 with a negligible amount of ferrite. (Table 3). Under static conditions Fe2 O3 is the equilibrium phase that forms under ambient conditions [16]. This is consistent with the measured bulk temperature of 60◦ C (Fig. 8) and the wear debris collected was a non-magnetic powder with a dark orange color indicative of hydrated Fe2 O3 . The wear rates were lower than those encountered at the same low velocity levels but at higher loads. 4.3. High velocities and low loads
Fig. 9. A schematic diagram illustrating a mechanism for deformation and fracture of steel splats: (a) shows a cross-section of the unworn surface; (b) shows a cross-section of plastically deformed splats elongated along the sliding direction; (c) is a surface view showing fracture at the edges of the splats.
At high velocity and low loads the wear rate had the lowest value compared to the other regimes (Fig. 2). The SEM micrograph (Fig. 5) of the worn surface of a sample tested under these conditions shows a thin deposit of oxide. However, coverage was more significant than at low load and low velocity conditions and extended over the top of the steel splats. The measured pin temperature for the sample tested at 10 N and 2.5 m/s was about 110◦ C, which was higher than the sample tested at low velocity and low load. Because, the wear rate was very low, insufficient wear debris could be collected for XRD investigation. It can be suggested that several favorable factors, including the formation of almost continuous oxide rich films, hardening of the coating during wear, and lack of any significant iron splat fracture and delamination, are among the possible reasons why the wear rates are the lowest under these conditions. 4.4. High velocities and high loads
into these regions much more easily than in dense smooth regions that are self-supporting. Another wear mechanism related to the lamellar structure of the coatings is the splat delamination. Oxide layers between splats are the weak links in many spray deposited coatings [7]. In [6], the high wear rates were generally associated with the formation and propagation of subsurface cracks within the oxide in the thermal spray deposited coatings, resulting in removal of complete steel splats during the sliding process. The splat orientation and surface waviness were suggested to have an influence on the delamination mechanism. Delamination was shown to be easier when the splats were parallel to the coating surface and more difficult when they were wavy and not parallel to interface [6]. Splat delamination was occasionally observed under the present experimental conditions but was not a dominant wear mechanism for the PTWA low carbon steel coatings. 4.2. Low loads and low velocities At low load and velocity oxidation appeared to be the main wear mechanism. Splat fracture was rarely observed in
At high velocities and high loads the wear rates are lower than at low velocities (Fig. 2). In fact the wear rates were about 50% lower than those measured at similar loads but using low test speeds, which sounds counter intuitive. However, the experimental data suggested that high surface temperatures played an important role to keep the wear rates low under these most severe testing conditions. The high surface temperatures may have reduced wear rates in two ways, (i) by producing thick oxide rich layers (Fig. 4) that reduce the coefficient of friction and probably suppressing the splat tip fracture mechanism, (ii) by increasing the hardness of the steel splat surfaces during wear (Fig. 7). The pin temperature (Fig. 8) shows that the sliding induced temperature rise was influenced more by sliding velocity than by load. The maximum bulk temperature rise occurred for the sample tested at 75 N and 2.5 m/s, where the increase was about 390◦ C. The wear track of the high velocity and high load sample had regions (about 70%) that were covered with a thick oxide rich layer whose thickness reached between 1 and 3 m at some points. Fig. 8 shows that by increasing velocity from 0.5 to 2.5 m/s at 50 N load the coefficient
A. Edrisy et al. / Wear 251 (2001) 1023–1033
of friction decreased from approximately 0.8 to 0. 6 this is also consistent with the formation of thick oxide rich film. The XRD spectrum of the wear debris of the sample tested at 50 N load and 2 m/s in Fig. 6(c) showed the presence of all three oxides. Although these phases were the same as those of the debris from low velocity and high load tests, the quantitative phase analysis shows the relative amount of ferrite at high velocity was lower. Evidence from SEM and OM suggested that the splat tip fracture mechanism that was responsible for the high wear rates at low load and velocity conditions was not as significant when the sliding speed was high. The hardness of some parts of the wear tracks in samples tested at high loads, e.g. 50 and 75 N and 2.5 m/s sliding velocity increased to over 800 kg/mm2 , from the initial 310 kg/mm2 for the unworn surface. Welsh attributed the excessive hardening of plain carbon steels during dry sliding to a martensitic phase transformation induced by temperature fluctuations on the sliding surfaces. It was also mentioned that when the temperature was high enough to allow an austenite transformation, the probability of absorption of nitrogen or the other constituents of the atmosphere was high [8,9]. The origin of this hardening effect in the PTWA low carbon steel coatings was not revealed by the wear experiments. The temperature should exceed 860◦ C for any martensite to form in 1020 steel by a conventional heating and quenching mechanism. In an attempt to simulate the effect of flash temperature at the asperities, PTWA coatings were instantaneously heated by electrical resistance (spark) heating and self-quenched in a bell jar under argon atmosphere. The resulting microhardness values were similar to those generated by sliding wear (750 kg/ mm2 ). The effect of surface hardening during wear should be studied in more detail. 5. Conclusions The wear rates of thermal spray deposited low carbon steel coatings produced by a PTWA process were measured at different load and velocity conditions. The wear mechanisms of the coating were associated with surface oxidation, splat tip fracture and surface hardening during wear. Splat delamination or removal of individual steel splats as a result of crack growth along the FeO veins within the coating was seldom observed. The wear rates and mechanisms can be classified as follows: 1. At low load and velocity, oxidation (formation of Fe2 O3 ) was the main wear mechanism.
1033
2. At low velocity and high load the high wear rate was associated with severe deformation of the steel splat tips and eventually splat fracture and fragmentation. In addition, tribological layers consisting of a mixture of Fe2 O3 , FeO, and Fe3 O4 formed on the contact surfaces. 3. At high load and velocity the wear rates decreased. Evidence was found for two different mechanisms that could account for the relatively low wear rates. The first mechanism was the formation of a thick protective mixed oxide layer on the wear track. The second mechanism was a pronounced hardening of the iron coating. 4. The wear rates were the lowest at high velocity and low load conditions, where there was no evidence for the splat tip fracture mechanism and the surfaces were covered with oxide rich tribolayers.
Acknowledgements The authors would like to thank Michael Balogh for his expertise with Rietveld analysis, T.W. Capehart for reading the manuscript, and John Robinson for his assistance with metallography and material characterization techniques. References [1] S. Wilson, A.T. Alpas, Wear 212 (1997) 41–49. [2] C.C. Berndt, S. Sampath (Eds.), Thermal Spray Industrial Applications, ASM International, Materials Park, OH, 1994. [3] L. Pawlowski, P. Fauchais, Int. Mater. Rev. 37 (6) (1992) 271–289. [4] V.V. Sobolev, J.M. Guilemany, J. Nutting, J.R. Miquel, Int. Mater. Rev. 42 (3) (1997) 117–136. [5] B.S. Schorr, K.J. Stein, A.R. Marder, Mater. Character. 42 (1999) 93–100. [6] S.E. Hartfield-Wunsch, S.C. Tung, in: C.C. Berndt, S. Sampath (Eds.), Thermal Spray Industrial Applications, ASM International, Materials Park, OH, 1994, pp. 19–32. [7] A. Rabiei, D.R. Mumm, J.W. Hutchinson, R. Schweinfest, M. Ruhle, A.G. Evans, Mater. Sci. Eng. A 269 (1999) 152–165. [8] N.C. Welsh, Royal Society of London, Philos. Trans. Series A 257 (1964) 34–52. [9] N.C. Welsh, J. Appl. Phys. 28 (9) (1957) 960–968. [10] J.F. Archard, Wear 2 (1952) 438. [11] S.C. Lim, M.F. Ashby, Acta Metallurgica. 35 (1987) 1–24. [12] T.F.J. Quinn, Tribol. Int. 16 (5) (1983) 257–271. [13] M.F. Ashby, J. Abulawi, H.S. Kong, Tribol. Trans. 34 (1991) 577–587. [14] I.M. Hutchings, Tribology, Friction and Wear of Engineering Materials, CRC Press, Cambridge, 1992, pp. 103–106. [15] T.F.J. Quinn, Oxidational wear in friction, lubrication, and wear technology, ASM Handbook, Vol. 18, ASM International, Metals Park, OH, 1998, pp. 280–288. [16] A. Muan, E.F. Osborn, Phase Equilibria Among Oxides in Steelmaking, Addison-Wesley, Reading, MA, 1965, p. 27.