- Email: [email protected]
Rolling contact fatigue of alumina ceramics sprayed on steel roller under pure rolling contact condition
ARTICLE IN PRESS
Tribology International 39 (2006) 856–862 www.elsevier.com/locate/triboint
Rolling contact fatigue of alumina ceramics sprayed on steel roller under pure rolling contact condition Masahiro Fujiia,, Akira Yoshidaa, Jiabin Mab, Sadato Shigemurac, Kazumi Tanic a
Department of Mechanical Engineering, Okayama University, Tsushima-Naka, Okayama 700-8530, Japan The Graduate School of Natural Science and Technology, Okayama University, Tsushima-Naka, Okayama 700-8530, Japan c Tocalo Co., Ltd., Fukae-Kitamachi, Higashinada-ku, Kobe 658-0013, Japan
b
Received 14 October 2004; received in revised form 20 June 2005; accepted 12 July 2005 Available online 10 October 2005
Abstract The rolling contact fatigue of sprayed alumina ceramics with a nominal composition of Al2O3–2.3 mass% TiO2 was studied with a two-roller test machine under a pure rolling contact condition with oil lubricant. The influence of undercoating of sprayed Ni-based alloy on the rolling contact fatigue was investigated. The failure mode of all sprayed rollers was spalling caused by subsurface cracking. The undercoating did not contribute to the improvement of the rolling contact fatigue life. The elastic modulus of the alumina sprayed layer evaluated with the nano-indentation method was around 85 GPa. The depths of the observed subsurface cracks corresponded approximately to the depths where the orthogonal shear stress or the maximum shear stress calculated with two-dimensional FEM became maximum. r 2005 Elsevier Ltd. All rights reserved. Keywords: Rolling contact fatigue; Spraying; Alumina ceramics; Oil lubrication; Nano-indentation; Stress analysis
1. Introduction Thermal spraying is one kind of surface modification treatment utilized in almost all industrial fields. The thickness of a sprayed layer ranges from several mm to dozens of mm. Various kinds of materials such as metals, ceramics and plastics can be used as spraying material. The alumina ceramics employed in this study have high hardness, superior chemical stability and high resistivity; thus, the alumina ceramics are frequently used when corrosion resistance, wear resistance and electric insulation are required. However, the fracture toughness of the alumina ceramics is low. Hence, there are hardly any applications to machine elements under concentrated load conditions such as point and line contacts. On the other hand, when sprayed on the surface of materials such as steel, which has a high fracture toughness value, the alumina ceramics are expected to be applicable under comparatively heavy loads as well as to improve the wear Corresponding author. Tel.: +81 86 251 8035; fax: +81 86 251 8266.
E-mail address: [email protected] (M. Fujii). 0301-679X/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.triboint.2005.07.038
and corrosion resistance of the surface. In order to apply alumina ceramics spraying to various rolls and some other rolling contact machine elements, it would be very useful to acquire the knowledge about its rolling contact fatigue and surface damage. Although the rolling contact fatigue of WC cermet and WC-Co sprayed layer has been studied [1–3], reports on mechanical properties and rolling contact fatigue of the sprayed layer of alumina ceramics are few. In this study, the rolling contact fatigue of alumina ceramics sprayed on steel rollers was investigated under a pure rolling contact condition with oil lubricant. Moreover, the modulus of elasticity of the sprayed layers was evaluated using a thin film hardness tester based on the nano-indentation method. The failure mode was also discussed in terms of subsurface stresses. 2. Test rollers and spraying condition Fig. 1 shows the shapes and dimensions of the test roller and the mating roller. The rollers were made of bearing steel SUJ2 (JIS) and 0.45% carbon steel S45C (JIS). The test rollers with a contact width of 8 mm were sprayed with
ARTICLE IN PRESS M. Fujii et al. / Tribology International 39 (2006) 856–862
857
Table 2 Test roller Specimen
BSR1 BSR2 CSR2
Roller material
SUJ2 SUJ2 S45C
Thickness
1.2 Al2O3 spraying (mm) Ni-based alloy undercoating (mm) —
1.0 0.2
1.0 0.2
Fig. 1. Shapes and dimensions of roller specimens. Table 1 Spraying condition Alumina ceramics (plasma spraying) Arc current (A) Arc voltage (V) Plasma gas flow rate (l/min) Primary (Ar) Secondary (H2) Spraying distance (mm)
41 10 120
Ni-based alloy undercoating (HVOF) Oxygen (l/min) Kerosene (l/min) Combustion pressure (MPa) Spraying distance (mm)
16 0.75 0.61 380
600 65
alumina ceramics. The mating rollers without the sprayed layer have a width of 10 mm. The SUJ2 rollers were quenched in oil after being heated to 1123 K and were tempered at 433 K. The S45C rollers were quenched in water after being heated to 1123 K and were tempered at 873 K. For the SUJ2 rollers, alumina ceramics were sprayed on their surface directly (hereinafter called BSR1) or after a nickel-based alloy undercoating (hereinafter called BSR2). For the S45C rollers, alumina ceramics were sprayed after a nickel-based alloy undercoating (hereinafter called CSR2). The spraying condition of alumina ceramics and undercoating condition of nickel-based alloy are shown in Table 1. The three kinds of sprayed rollers used in this study are listed in Table 2. Alumina ceramics with a nominal composition of Al2O3–2.3 mass% TiO2 were deposited with an atmospheric plasma spraying process. The thickness of the layer laminated by each spraying process was about 20 mm. The nickel-based alloy was deposited with a high-velocity oxygen fuel (HVOF) process for the purpose of fine microstructure of the sprayed layer. The diameters of the sprayed particles of both the alumina ceramics and the nickel-based alloy were 10–45 mm. All the sprayed rollers were finished by grinding. For the sprayed rollers without an undercoating, the thickness of the sprayed layer of alumina ceramics was 1.2 mm. For the sprayed rollers
Fig. 2. Observation of sprayed layer.
with a nickel-based alloy undercoating, the thicknesses of alumina ceramics and nickel-based alloy layers were1.0 and 0.2 mm, respectively. With a view to reducing the influence of the difference of thermal expansion between the matrix and the sprayed layer, the practically sufficient thickness of the nickel-based alloy undercoating was 0.05–0.075 mm, where the microstructure of particle laminations was almost continuous. In this study, however, the thickness of the sprayed nickel-based alloy layer was 0.2 mm, with additional consideration of environmental insulation. The surface of the sprayed roller was sealed with epoxy resin to prevent the infiltration of lubricant into surface pores. The arithmetic surface roughnesses of the sprayed roller and the mating roller were about 0.37 and 0.31–0.43 mm Ra, respectively. The average hardnesses of the sprayed layer and the matrix (SUJ2 steel), which were measured with a thin film hardness tester that was used for evaluation of elastic modulus mentioned later, were HUT 406 and HUT 225 kg f/mm2, respectively. Fig. 2 shows the surface observation of sprayed layer. The average porosity was around 26%. 3. Test method The rolling fatigue test was performed under a pure rolling contact condition with a spring loading type tworoller test machine shown in Fig. 3. Each roller was driven with an individual electric motor, and the rotation speed of both rollers was 2000 rpm. The average rolling velocity of the rollers was 4.2 m/s. The lubricating oil was #83 turbine oil. The testing load per unit width W/L (W—load, L— contact width) was 100–200 N/mm. An acceleration pickup
ARTICLE IN PRESS M. Fujii et al. / Tribology International 39 (2006) 856–862
858
250 BSR2 BSR2 BSR1
W/L, N/mm
200
150
Fig. 3. Schema of two-roller test machine.
fixed on the bearing box near the tested roller was adjusted to stop automatically in response to the surface failure of the test roller. The number of rotations of the sprayed roller at this automatic stop was taken as the rolling contact fatigue life. In addition, when the number of cycles of the sprayed roller amounted to 107 times without failure, the test was stopped.
100 105
106 Number of cycles to failure N
107
Fig. 4. W/L– N curves.
4. Test results and discussion 4.1. Rolling contact fatigue Fig. 4 shows the relation between load per unit width W/ L and rolling contact fatigue life N of the sprayed rollers. Fig. 5 shows surface observations of failed rollers. As shown in Fig. 5, in most cases, the surface failure extended to a wide area when the test machine stopped automatically. This surface failure may occur within a short time. However, there was no roller whose whole sprayed layer peeled off. The spalling failure occurred in the sprayed layer as described later. There was little difference between the rolling contact fatigue lives of BSR2 rollers made of SUJ2 and of CSR rollers made of S45C. This result indicated that there was no significant influence of the matrix steel of the roller on the rolling contact fatigue of the sprayed layer when the spalling failure occurred in the sprayed layer. Although the rolling contact fatigue strength of BSR1 roller without undercoating could be slightly higher than that of BSR2 roller with undercoating, the influence of the undercoating on the rolling contact fatigue was little. Generally, even if the nominal spraying conditions are the same, the small variation of gas pressure and temperature in the spraying process could lead to a variation in the size and distribution of pores and the mechanical strength of the sprayed layer, which greatly affects the initiation and propagation of cracks. The variation of the rolling contact fatigue strength of fine ceramics rollers was larger than that of steel rollers, which was shown by the authors’ study using bulk ceramics rollers [4]. In addition, the rolling contact fatigue strength of the alumina ceramics sprayed on the roller would vary more widely than that of the bulk ceramics roller, because the mechanical properties of the sprayed layer could vary
Fig. 5. Observation of failed roller.
with spraying conditions. Therefore, the small difference in the rolling contact fatigue strength shown in Fig. 4 would be due to the inevitable variation of the mechanical properties of sprayed layers. Fig. 6 shows photographs of cross-sections perpendicular to the axis of the failed sprayed rollers. Subsurface cracks propagating in the circumferential direction parallel to the roller surface were observed. The depths of these subsurface cracks were about 70 mm in BSR1 and about 200 mm in BSR2. The failure mode was spalling caused by subsurface cracking. Some cracks extending from the roller surface to the subsurface circumferential crack in the radial direction were also observed in Fig. 6(b). The subsurface circumferential crack, i.e., the spalling crack, would occur prior to the radial cracks. In some other failed rollers, the
ARTICLE IN PRESS M. Fujii et al. / Tribology International 39 (2006) 856–862
Fig. 6. Cross-section of failed roller.
859
seen in the interface between the sprayed layer and the matrix, but it did not progress to a large peeling. For the BSR2 roller shown in Fig. 7(b), a crack occurred in the sprayed layer near the interface between the nickel-based alloy undercoating and the alumina ceramics sprayed layer. Although the undercoating improved the bonding strength between the alumina sprayed layer and the matrix, it did not always contribute to the improvement of the rolling contact fatigue strength as shown in Fig. 4. When the elastic modulus of the sprayed layer is different from that of the matrix, the shear stress distribution below the sprayed roller surface under rolling contact is discontinuous near the interface between the sprayed layer and the matrix. As the linear thermal expansion coefficient of the sprayed layer is different from that of the matrix, the residual stress would occur near the interface during the cooling process from a relatively high spraying temperature. Although the discontinuous shear stress and residual stress distributions near the interface would be moderated by the nickel-based alloy undercoating, these would cause cracks as observed in the interface between the sprayed layer and the matrix. 4.2. Elastic modulus of sprayed layer The hardness and elastic modulus of the thin film has been obtained using the relation between load P and indentation depth h measured with the nano-indentation tester [5,6]. The stiffness S obtained from the slope of unloading curve is given by pffiffiffiffi 2 S ¼ pffiffiffi E r A, (1) p where A is contact area of indenter and Er is reduced elastic modulus given by 1 1 v2 1 v2i þ ¼ , Er E Ei
(2)
where E and n are elastic modulus and Poisson’s ratio of the tested material, and Ei and vi are elastic modulus and Poisson’s ratio of the indenter. For a diamond indenter, Ei is 1141 GPa and ni is 0.07. In this study, the diamond indenter whose shape is the trigonal pyramid with a face angle of 681 to the axis was employed. The relation between contact area A and indentation depth h of the indenter is geometrically designated A ¼ 34:332h2 .
Fig. 7. Cross-section of failed roller.
subsurface spalling cracks were also observed at depths of 70–300 mm. Fig. 7 shows photographs of the interface between the sprayed layer and the matrix. For the BSR1 roller without undercoating shown in Fig. 7(a), a crack was
(3)
However, the radius of curvature of the indenter tip is of the order of 10 nm and, furthermore, the indenter tip is worn by repeated testing. Therefore, in order to evaluate the precise value of elastic modulus and hardness, the calibration of the relation between the contact area and the indentation depth during the loading process has been performed with single crystalline silicon or fused silica [5,6]. Because the stiffness of the apparatus as well as the shape
ARTICLE IN PRESS M. Fujii et al. / Tribology International 39 (2006) 856–862
860
of the indenter tip affects the test results, it is also necessary to obtain the compliance of the apparatus [6]. There is no easy method of calibration and, moreover, the calibration must be carried out for each test. In this study, the elastic modulus of the sprayed layer was evaluated by the following procedure. The elastic moduli of the sprayed layer and the matrix were measured for each test and the ratio of their elastic moduli was evaluated. Here, the influence of the condition of the environment and the apparatus on the evaluated values was expected to be minimized. The contact area of the indenter was a function of the indentation depth as shown in formula (3); therefore, the influence of the shape of the indenter tip was also reduced by controlling the indentation depth to be the same depth. The formula P ¼ aðh bÞm , where P is load, h is displacement, and a, b and m are constants, was applied to the load–displacement curve in the range of
Load P, mN
400
200
0
1 Displacement h, µm
2
SUJ2
(a) 600
Load P, mN
400
65–100% of the maximum load Pmax with regression analysis, and the stiffness S was obtained by the formula S ¼ amðh bÞm1 [6]. Fig. 8 shows the relation between load and indentation depth. The measurement was carried out with indentation depths of 1 and 2 mm five times each. The variation in the sprayed layer was larger than that in the matrix of SUJ2 steel. Fig. 9 shows an example of the residual indentation on the sprayed layer. The indenter was penetrated into pores in the sprayed layer. This would be a cause of the variation in the load–displacement curves shown in Fig. 8. There was difficulty in evaluating the Poisson’s ratio of the sprayed layer. Assuming that the Poisson’s ratio of the sprayed layer was 0.3, the ratios of the elastic modulus of the sprayed layer to that of the matrix were 0.43 for an indentation depth of 1 mm and 0.41 for an indentation depth of 2 mm. As the elastic modulus of SUJ2 steel was 206 GPa, the evaluated elastic modulus of the sprayed alumina ceramics layer was around 85 GPa. Generally, the elastic modulus of the alumina ceramics would be around 350 GPa, while the elastic modulus evaluated with this method was smaller than that value. As shown in Fig. 9, the indenter was penetrated into pores in the sprayed layer. It could be considered, therefore, that the indentation depth of the sprayed layer was deeper than that of bulk ceramics, and this resulted in a smaller value of the evaluated elastic modulus of the sprayed layer. 4.3. Subsurface shear stresses
200
0 (b)
Fig. 9. Photograph of residual indentation on sprayed layer.
1 Displacement h, µm Sprayed layer Fig. 8. Series of nano-indentation tests.
2
For ductile material such as steel, the spalling failure caused by subsurface cracking has been related to orthogonal shear stress and maximum shear stress [7]. On the other hand, the failure in brittle material like fine ceramics has been related to the principal stress. In the authors’ study on the rolling contact fatigue of fine ceramics rollers, the depth of spalling occurrence has been related to shear stresses [4]. Since the failure mode of the alumina ceramics sprayed on the roller was spalling caused by subsurface cracking in this study, the relation between shear stresses and spalling failure was discussed using the
ARTICLE IN PRESS M. Fujii et al. / Tribology International 39 (2006) 856–862
results obtained by two-dimensional FEM analysis. In this analysis, the elastic modulus Em and the Poisson’s ratio nm of the matrix were 206 GPa and 0.3, respectively, and the Poisson’s ratio ns of the sprayed layer was also 0.3. Although the elastic modulus of the sprayed layer evaluated in the previous section was around 85 GPa, here the shear stresses were analyzed with elastic moduli of the sprayed layer Es of 50–400 GPa involving that of bulk fine
Fig. 10. Distribution of orthogonal shear stress tyz and maximum shear stress tmax.
861
ceramics. The analysis was performed as a contact problem between roller and plane under the conditions of load per unit width, W/L of 100 and 200 N/mm. The relative radius of curvature corresponded with that of the test rollers. The y and z coordinates of the roller are established in the circumferential and the radial directions of the roller, respectively. Fig. 10 shows the distribution of the maximum shear stress tmax and the orthogonal shear stress tyz analyzed by FEM. The plotted values were the maximum values of shear stresses in each depth. The orthogonal shear stress tyz and the maximum shear stress tmax at a certain point below the contact surface change in alternating manner and in pulsating manner by the cyclic rolling contact, respectively. For both W =L ¼ 100 N=mm and W =L ¼ 200 N=mm, the depths at the maximum values of both the maximum shear stress tmax and the orthogonal shear stress tyz were within the sprayed layer. Moreover, as the elastic modulus of the sprayed layer became large, the maximum values of both the maximum shear stress tmax and the orthogonal shear stress tyz and the depths at the maximum values became large. For E s ¼ 85 GPa, the depths at the maximum of tmax and the maximum of tyz under W =L ¼ 100 N=mm were about 80 and 120 mm, respectively, and the depths under W =L ¼ 200 N=mm were about 120 and 180 mm, respectively. The depths of spalling cracks observed in this study ranged from 70 to 300 mm, which involved the depths at the maximum of shear stresses analyzed by FEM. In the analysis shown in Fig. 10, the sprayed layer was assumed to be isotropic. Since the sprayed layer is formed by the layered particles which are melted or semi-melted in the spraying process, pores and discontinuous laminations were formed in the sprayed layer. Therefore, it can be understood that the composition and the mechanical properties of the sprayed layer tend to be inhomogeneous, and the mechanical property of the sprayed layer in the circumferential direction of the roller is different from that in the radial direction. In addition, the sprayed ceramics layer with a number of pores has a fundamental variation in strength. Taking the above factors into consideration, the depths at the maximum of shear stresses would correspond to the depth of the spalling cracks. The maximum contact pressure of the spayed roller calculated with FEM analysis was compared with the maximum Hertzian pressure of a bulk ceramics roller. The difference was up to 3.5%. This suggests that the shear stresses in the sprayed layer could be evaluated with the theoretical solution by Smith et al., assuming that the roller was made of bulk ceramics [8]. Fig. 11 shows the distribution of orthogonal shear stresses calculated with FEM analysis and the method by Smith et al. Because the method by Smith et al. cannot deal with the case of a roller coated with a sprayed layer whose mechanical property is different from that of the matrix, the calculation was performed assuming that the roller was made of bulk ceramics. For FEM analysis, a slight discontinuity of the shear stress distribution was observed near the interface
ARTICLE IN PRESS M. Fujii et al. / Tribology International 39 (2006) 856–862
862
conditions, and the relation between the failure mode and the subsurface shear stresses was discussed. The results obtained are as follows: (1) The failure mode of the sprayed rollers was spalling caused by subsurface cracking. (2) When the failure depth was smaller compared with the thickness of the sprayed layer, the effect of undercoating on the rolling contact fatigue strength was little. (3) The elastic modulus of the sprayed alumina ceramics layer evaluated with the nano-indentation method was around 85 GPa. (4) Assuming the sprayed layer to be uniform and isotropic, the depths at the maximum values of subsurface shear stresses were within the sprayed layer and almost corresponded to the depths of the observed subsurface cracks.
Acknowledgement Fig. 11. Distribution of orthogonal shear stress tyz.
between the sprayed layer and the matrix. Except for the interface, the orthogonal shear stress by FEM agrees with that by the theoretical solution. When the thickness of the sprayed layer is comparatively thick and the depth at the maximum of shear stress is comparatively shallow, shear stresses could be evaluated with the theoretical solution assuming that the roller was made of bulk ceramics. In this study, there was little difference between the rolling contact fatigue lives of the sprayed rollers with an undercoating and of those without an undercoating. This would be due to the fact that the values of shear stresses near the interface between the sprayed layer and the matrix were not always great. The thickness of the sprayed layer and the necessity of undercoating should be designed, taking the depth where the maximum value of shear stress occurs into consideration. 5. Conclusions The rolling contact fatigue of alumina ceramics sprayed on the rollers was investigated under oil lubricating
The authors wish to express their gratitude to Mr. Kazuhiko Hagiwara for his helpful assistance. References [1] Ahmen R, Hadfield M. Rolling contact fatigue performance of plasma sprayed coatings. Wear 1998;220(1):80–91. [2] Nieminen R, Vuoristo P, Niemi K, Ma¨ntyla¨ T, Barbezat G. Rolling contact fatigue failure mechanisms in plasma and HVOF sprayed WCCo coatings. Wear 1997;212(1):66–77. [3] Yoshida M, Tani K, Nakahira A, Nakajima A, Mawatari T. Durability and tribological properties of thermally sprayed WC cermet coating in rolling/sliding contact. Proceedings of ITSC, 1995, Kobe. 1995. p. 663–668. [4] Yoshida A, Fujii M, Nagamori K, Haishi H, Uchimoto H. Study on rolling contact fatigue of fine ceramics (part 4: in a case of pure rolling contact against steel). Jpn J Tribol 1992;37(1):63–74. [5] Doerner MF, Nix WD. A method for interpreting the data from depth-sensing indentation instruments. J Mater Res 1986;1(4):601–9. [6] Oliver WC, Pharr GM. An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments. J Mater Res 1992;7(6):1564–83. [7] Fujita K, Yoshida A. Surface durability of steel rollers (in the case of case-hardened and of nitrided rollers). Bull JSME 1978;21(154):761–7. [8] Smith JO, Liu CK. Stresses due to tangential and normal loads on an elastic solid with application to some contact stress problems. Trans. ASME, J Appl Mech 1953;20:157–66.
Tribology International 39 (2006) 856–862 www.elsevier.com/locate/triboint
Rolling contact fatigue of alumina ceramics sprayed on steel roller under pure rolling contact condition Masahiro Fujiia,, Akira Yoshidaa, Jiabin Mab, Sadato Shigemurac, Kazumi Tanic a
Department of Mechanical Engineering, Okayama University, Tsushima-Naka, Okayama 700-8530, Japan The Graduate School of Natural Science and Technology, Okayama University, Tsushima-Naka, Okayama 700-8530, Japan c Tocalo Co., Ltd., Fukae-Kitamachi, Higashinada-ku, Kobe 658-0013, Japan
b
Received 14 October 2004; received in revised form 20 June 2005; accepted 12 July 2005 Available online 10 October 2005
Abstract The rolling contact fatigue of sprayed alumina ceramics with a nominal composition of Al2O3–2.3 mass% TiO2 was studied with a two-roller test machine under a pure rolling contact condition with oil lubricant. The influence of undercoating of sprayed Ni-based alloy on the rolling contact fatigue was investigated. The failure mode of all sprayed rollers was spalling caused by subsurface cracking. The undercoating did not contribute to the improvement of the rolling contact fatigue life. The elastic modulus of the alumina sprayed layer evaluated with the nano-indentation method was around 85 GPa. The depths of the observed subsurface cracks corresponded approximately to the depths where the orthogonal shear stress or the maximum shear stress calculated with two-dimensional FEM became maximum. r 2005 Elsevier Ltd. All rights reserved. Keywords: Rolling contact fatigue; Spraying; Alumina ceramics; Oil lubrication; Nano-indentation; Stress analysis
1. Introduction Thermal spraying is one kind of surface modification treatment utilized in almost all industrial fields. The thickness of a sprayed layer ranges from several mm to dozens of mm. Various kinds of materials such as metals, ceramics and plastics can be used as spraying material. The alumina ceramics employed in this study have high hardness, superior chemical stability and high resistivity; thus, the alumina ceramics are frequently used when corrosion resistance, wear resistance and electric insulation are required. However, the fracture toughness of the alumina ceramics is low. Hence, there are hardly any applications to machine elements under concentrated load conditions such as point and line contacts. On the other hand, when sprayed on the surface of materials such as steel, which has a high fracture toughness value, the alumina ceramics are expected to be applicable under comparatively heavy loads as well as to improve the wear Corresponding author. Tel.: +81 86 251 8035; fax: +81 86 251 8266.
E-mail address: [email protected] (M. Fujii). 0301-679X/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.triboint.2005.07.038
and corrosion resistance of the surface. In order to apply alumina ceramics spraying to various rolls and some other rolling contact machine elements, it would be very useful to acquire the knowledge about its rolling contact fatigue and surface damage. Although the rolling contact fatigue of WC cermet and WC-Co sprayed layer has been studied [1–3], reports on mechanical properties and rolling contact fatigue of the sprayed layer of alumina ceramics are few. In this study, the rolling contact fatigue of alumina ceramics sprayed on steel rollers was investigated under a pure rolling contact condition with oil lubricant. Moreover, the modulus of elasticity of the sprayed layers was evaluated using a thin film hardness tester based on the nano-indentation method. The failure mode was also discussed in terms of subsurface stresses. 2. Test rollers and spraying condition Fig. 1 shows the shapes and dimensions of the test roller and the mating roller. The rollers were made of bearing steel SUJ2 (JIS) and 0.45% carbon steel S45C (JIS). The test rollers with a contact width of 8 mm were sprayed with
ARTICLE IN PRESS M. Fujii et al. / Tribology International 39 (2006) 856–862
857
Table 2 Test roller Specimen
BSR1 BSR2 CSR2
Roller material
SUJ2 SUJ2 S45C
Thickness
1.2 Al2O3 spraying (mm) Ni-based alloy undercoating (mm) —
1.0 0.2
1.0 0.2
Fig. 1. Shapes and dimensions of roller specimens. Table 1 Spraying condition Alumina ceramics (plasma spraying) Arc current (A) Arc voltage (V) Plasma gas flow rate (l/min) Primary (Ar) Secondary (H2) Spraying distance (mm)
41 10 120
Ni-based alloy undercoating (HVOF) Oxygen (l/min) Kerosene (l/min) Combustion pressure (MPa) Spraying distance (mm)
16 0.75 0.61 380
600 65
alumina ceramics. The mating rollers without the sprayed layer have a width of 10 mm. The SUJ2 rollers were quenched in oil after being heated to 1123 K and were tempered at 433 K. The S45C rollers were quenched in water after being heated to 1123 K and were tempered at 873 K. For the SUJ2 rollers, alumina ceramics were sprayed on their surface directly (hereinafter called BSR1) or after a nickel-based alloy undercoating (hereinafter called BSR2). For the S45C rollers, alumina ceramics were sprayed after a nickel-based alloy undercoating (hereinafter called CSR2). The spraying condition of alumina ceramics and undercoating condition of nickel-based alloy are shown in Table 1. The three kinds of sprayed rollers used in this study are listed in Table 2. Alumina ceramics with a nominal composition of Al2O3–2.3 mass% TiO2 were deposited with an atmospheric plasma spraying process. The thickness of the layer laminated by each spraying process was about 20 mm. The nickel-based alloy was deposited with a high-velocity oxygen fuel (HVOF) process for the purpose of fine microstructure of the sprayed layer. The diameters of the sprayed particles of both the alumina ceramics and the nickel-based alloy were 10–45 mm. All the sprayed rollers were finished by grinding. For the sprayed rollers without an undercoating, the thickness of the sprayed layer of alumina ceramics was 1.2 mm. For the sprayed rollers
Fig. 2. Observation of sprayed layer.
with a nickel-based alloy undercoating, the thicknesses of alumina ceramics and nickel-based alloy layers were1.0 and 0.2 mm, respectively. With a view to reducing the influence of the difference of thermal expansion between the matrix and the sprayed layer, the practically sufficient thickness of the nickel-based alloy undercoating was 0.05–0.075 mm, where the microstructure of particle laminations was almost continuous. In this study, however, the thickness of the sprayed nickel-based alloy layer was 0.2 mm, with additional consideration of environmental insulation. The surface of the sprayed roller was sealed with epoxy resin to prevent the infiltration of lubricant into surface pores. The arithmetic surface roughnesses of the sprayed roller and the mating roller were about 0.37 and 0.31–0.43 mm Ra, respectively. The average hardnesses of the sprayed layer and the matrix (SUJ2 steel), which were measured with a thin film hardness tester that was used for evaluation of elastic modulus mentioned later, were HUT 406 and HUT 225 kg f/mm2, respectively. Fig. 2 shows the surface observation of sprayed layer. The average porosity was around 26%. 3. Test method The rolling fatigue test was performed under a pure rolling contact condition with a spring loading type tworoller test machine shown in Fig. 3. Each roller was driven with an individual electric motor, and the rotation speed of both rollers was 2000 rpm. The average rolling velocity of the rollers was 4.2 m/s. The lubricating oil was #83 turbine oil. The testing load per unit width W/L (W—load, L— contact width) was 100–200 N/mm. An acceleration pickup
ARTICLE IN PRESS M. Fujii et al. / Tribology International 39 (2006) 856–862
858
250 BSR2 BSR2 BSR1
W/L, N/mm
200
150
Fig. 3. Schema of two-roller test machine.
fixed on the bearing box near the tested roller was adjusted to stop automatically in response to the surface failure of the test roller. The number of rotations of the sprayed roller at this automatic stop was taken as the rolling contact fatigue life. In addition, when the number of cycles of the sprayed roller amounted to 107 times without failure, the test was stopped.
100 105
106 Number of cycles to failure N
107
Fig. 4. W/L– N curves.
4. Test results and discussion 4.1. Rolling contact fatigue Fig. 4 shows the relation between load per unit width W/ L and rolling contact fatigue life N of the sprayed rollers. Fig. 5 shows surface observations of failed rollers. As shown in Fig. 5, in most cases, the surface failure extended to a wide area when the test machine stopped automatically. This surface failure may occur within a short time. However, there was no roller whose whole sprayed layer peeled off. The spalling failure occurred in the sprayed layer as described later. There was little difference between the rolling contact fatigue lives of BSR2 rollers made of SUJ2 and of CSR rollers made of S45C. This result indicated that there was no significant influence of the matrix steel of the roller on the rolling contact fatigue of the sprayed layer when the spalling failure occurred in the sprayed layer. Although the rolling contact fatigue strength of BSR1 roller without undercoating could be slightly higher than that of BSR2 roller with undercoating, the influence of the undercoating on the rolling contact fatigue was little. Generally, even if the nominal spraying conditions are the same, the small variation of gas pressure and temperature in the spraying process could lead to a variation in the size and distribution of pores and the mechanical strength of the sprayed layer, which greatly affects the initiation and propagation of cracks. The variation of the rolling contact fatigue strength of fine ceramics rollers was larger than that of steel rollers, which was shown by the authors’ study using bulk ceramics rollers [4]. In addition, the rolling contact fatigue strength of the alumina ceramics sprayed on the roller would vary more widely than that of the bulk ceramics roller, because the mechanical properties of the sprayed layer could vary
Fig. 5. Observation of failed roller.
with spraying conditions. Therefore, the small difference in the rolling contact fatigue strength shown in Fig. 4 would be due to the inevitable variation of the mechanical properties of sprayed layers. Fig. 6 shows photographs of cross-sections perpendicular to the axis of the failed sprayed rollers. Subsurface cracks propagating in the circumferential direction parallel to the roller surface were observed. The depths of these subsurface cracks were about 70 mm in BSR1 and about 200 mm in BSR2. The failure mode was spalling caused by subsurface cracking. Some cracks extending from the roller surface to the subsurface circumferential crack in the radial direction were also observed in Fig. 6(b). The subsurface circumferential crack, i.e., the spalling crack, would occur prior to the radial cracks. In some other failed rollers, the
ARTICLE IN PRESS M. Fujii et al. / Tribology International 39 (2006) 856–862
Fig. 6. Cross-section of failed roller.
859
seen in the interface between the sprayed layer and the matrix, but it did not progress to a large peeling. For the BSR2 roller shown in Fig. 7(b), a crack occurred in the sprayed layer near the interface between the nickel-based alloy undercoating and the alumina ceramics sprayed layer. Although the undercoating improved the bonding strength between the alumina sprayed layer and the matrix, it did not always contribute to the improvement of the rolling contact fatigue strength as shown in Fig. 4. When the elastic modulus of the sprayed layer is different from that of the matrix, the shear stress distribution below the sprayed roller surface under rolling contact is discontinuous near the interface between the sprayed layer and the matrix. As the linear thermal expansion coefficient of the sprayed layer is different from that of the matrix, the residual stress would occur near the interface during the cooling process from a relatively high spraying temperature. Although the discontinuous shear stress and residual stress distributions near the interface would be moderated by the nickel-based alloy undercoating, these would cause cracks as observed in the interface between the sprayed layer and the matrix. 4.2. Elastic modulus of sprayed layer The hardness and elastic modulus of the thin film has been obtained using the relation between load P and indentation depth h measured with the nano-indentation tester [5,6]. The stiffness S obtained from the slope of unloading curve is given by pffiffiffiffi 2 S ¼ pffiffiffi E r A, (1) p where A is contact area of indenter and Er is reduced elastic modulus given by 1 1 v2 1 v2i þ ¼ , Er E Ei
(2)
where E and n are elastic modulus and Poisson’s ratio of the tested material, and Ei and vi are elastic modulus and Poisson’s ratio of the indenter. For a diamond indenter, Ei is 1141 GPa and ni is 0.07. In this study, the diamond indenter whose shape is the trigonal pyramid with a face angle of 681 to the axis was employed. The relation between contact area A and indentation depth h of the indenter is geometrically designated A ¼ 34:332h2 .
Fig. 7. Cross-section of failed roller.
subsurface spalling cracks were also observed at depths of 70–300 mm. Fig. 7 shows photographs of the interface between the sprayed layer and the matrix. For the BSR1 roller without undercoating shown in Fig. 7(a), a crack was
(3)
However, the radius of curvature of the indenter tip is of the order of 10 nm and, furthermore, the indenter tip is worn by repeated testing. Therefore, in order to evaluate the precise value of elastic modulus and hardness, the calibration of the relation between the contact area and the indentation depth during the loading process has been performed with single crystalline silicon or fused silica [5,6]. Because the stiffness of the apparatus as well as the shape
ARTICLE IN PRESS M. Fujii et al. / Tribology International 39 (2006) 856–862
860
of the indenter tip affects the test results, it is also necessary to obtain the compliance of the apparatus [6]. There is no easy method of calibration and, moreover, the calibration must be carried out for each test. In this study, the elastic modulus of the sprayed layer was evaluated by the following procedure. The elastic moduli of the sprayed layer and the matrix were measured for each test and the ratio of their elastic moduli was evaluated. Here, the influence of the condition of the environment and the apparatus on the evaluated values was expected to be minimized. The contact area of the indenter was a function of the indentation depth as shown in formula (3); therefore, the influence of the shape of the indenter tip was also reduced by controlling the indentation depth to be the same depth. The formula P ¼ aðh bÞm , where P is load, h is displacement, and a, b and m are constants, was applied to the load–displacement curve in the range of
Load P, mN
400
200
0
1 Displacement h, µm
2
SUJ2
(a) 600
Load P, mN
400
65–100% of the maximum load Pmax with regression analysis, and the stiffness S was obtained by the formula S ¼ amðh bÞm1 [6]. Fig. 8 shows the relation between load and indentation depth. The measurement was carried out with indentation depths of 1 and 2 mm five times each. The variation in the sprayed layer was larger than that in the matrix of SUJ2 steel. Fig. 9 shows an example of the residual indentation on the sprayed layer. The indenter was penetrated into pores in the sprayed layer. This would be a cause of the variation in the load–displacement curves shown in Fig. 8. There was difficulty in evaluating the Poisson’s ratio of the sprayed layer. Assuming that the Poisson’s ratio of the sprayed layer was 0.3, the ratios of the elastic modulus of the sprayed layer to that of the matrix were 0.43 for an indentation depth of 1 mm and 0.41 for an indentation depth of 2 mm. As the elastic modulus of SUJ2 steel was 206 GPa, the evaluated elastic modulus of the sprayed alumina ceramics layer was around 85 GPa. Generally, the elastic modulus of the alumina ceramics would be around 350 GPa, while the elastic modulus evaluated with this method was smaller than that value. As shown in Fig. 9, the indenter was penetrated into pores in the sprayed layer. It could be considered, therefore, that the indentation depth of the sprayed layer was deeper than that of bulk ceramics, and this resulted in a smaller value of the evaluated elastic modulus of the sprayed layer. 4.3. Subsurface shear stresses
200
0 (b)
Fig. 9. Photograph of residual indentation on sprayed layer.
1 Displacement h, µm Sprayed layer Fig. 8. Series of nano-indentation tests.
2
For ductile material such as steel, the spalling failure caused by subsurface cracking has been related to orthogonal shear stress and maximum shear stress [7]. On the other hand, the failure in brittle material like fine ceramics has been related to the principal stress. In the authors’ study on the rolling contact fatigue of fine ceramics rollers, the depth of spalling occurrence has been related to shear stresses [4]. Since the failure mode of the alumina ceramics sprayed on the roller was spalling caused by subsurface cracking in this study, the relation between shear stresses and spalling failure was discussed using the
ARTICLE IN PRESS M. Fujii et al. / Tribology International 39 (2006) 856–862
results obtained by two-dimensional FEM analysis. In this analysis, the elastic modulus Em and the Poisson’s ratio nm of the matrix were 206 GPa and 0.3, respectively, and the Poisson’s ratio ns of the sprayed layer was also 0.3. Although the elastic modulus of the sprayed layer evaluated in the previous section was around 85 GPa, here the shear stresses were analyzed with elastic moduli of the sprayed layer Es of 50–400 GPa involving that of bulk fine
Fig. 10. Distribution of orthogonal shear stress tyz and maximum shear stress tmax.
861
ceramics. The analysis was performed as a contact problem between roller and plane under the conditions of load per unit width, W/L of 100 and 200 N/mm. The relative radius of curvature corresponded with that of the test rollers. The y and z coordinates of the roller are established in the circumferential and the radial directions of the roller, respectively. Fig. 10 shows the distribution of the maximum shear stress tmax and the orthogonal shear stress tyz analyzed by FEM. The plotted values were the maximum values of shear stresses in each depth. The orthogonal shear stress tyz and the maximum shear stress tmax at a certain point below the contact surface change in alternating manner and in pulsating manner by the cyclic rolling contact, respectively. For both W =L ¼ 100 N=mm and W =L ¼ 200 N=mm, the depths at the maximum values of both the maximum shear stress tmax and the orthogonal shear stress tyz were within the sprayed layer. Moreover, as the elastic modulus of the sprayed layer became large, the maximum values of both the maximum shear stress tmax and the orthogonal shear stress tyz and the depths at the maximum values became large. For E s ¼ 85 GPa, the depths at the maximum of tmax and the maximum of tyz under W =L ¼ 100 N=mm were about 80 and 120 mm, respectively, and the depths under W =L ¼ 200 N=mm were about 120 and 180 mm, respectively. The depths of spalling cracks observed in this study ranged from 70 to 300 mm, which involved the depths at the maximum of shear stresses analyzed by FEM. In the analysis shown in Fig. 10, the sprayed layer was assumed to be isotropic. Since the sprayed layer is formed by the layered particles which are melted or semi-melted in the spraying process, pores and discontinuous laminations were formed in the sprayed layer. Therefore, it can be understood that the composition and the mechanical properties of the sprayed layer tend to be inhomogeneous, and the mechanical property of the sprayed layer in the circumferential direction of the roller is different from that in the radial direction. In addition, the sprayed ceramics layer with a number of pores has a fundamental variation in strength. Taking the above factors into consideration, the depths at the maximum of shear stresses would correspond to the depth of the spalling cracks. The maximum contact pressure of the spayed roller calculated with FEM analysis was compared with the maximum Hertzian pressure of a bulk ceramics roller. The difference was up to 3.5%. This suggests that the shear stresses in the sprayed layer could be evaluated with the theoretical solution by Smith et al., assuming that the roller was made of bulk ceramics [8]. Fig. 11 shows the distribution of orthogonal shear stresses calculated with FEM analysis and the method by Smith et al. Because the method by Smith et al. cannot deal with the case of a roller coated with a sprayed layer whose mechanical property is different from that of the matrix, the calculation was performed assuming that the roller was made of bulk ceramics. For FEM analysis, a slight discontinuity of the shear stress distribution was observed near the interface
ARTICLE IN PRESS M. Fujii et al. / Tribology International 39 (2006) 856–862
862
conditions, and the relation between the failure mode and the subsurface shear stresses was discussed. The results obtained are as follows: (1) The failure mode of the sprayed rollers was spalling caused by subsurface cracking. (2) When the failure depth was smaller compared with the thickness of the sprayed layer, the effect of undercoating on the rolling contact fatigue strength was little. (3) The elastic modulus of the sprayed alumina ceramics layer evaluated with the nano-indentation method was around 85 GPa. (4) Assuming the sprayed layer to be uniform and isotropic, the depths at the maximum values of subsurface shear stresses were within the sprayed layer and almost corresponded to the depths of the observed subsurface cracks.
Acknowledgement Fig. 11. Distribution of orthogonal shear stress tyz.
between the sprayed layer and the matrix. Except for the interface, the orthogonal shear stress by FEM agrees with that by the theoretical solution. When the thickness of the sprayed layer is comparatively thick and the depth at the maximum of shear stress is comparatively shallow, shear stresses could be evaluated with the theoretical solution assuming that the roller was made of bulk ceramics. In this study, there was little difference between the rolling contact fatigue lives of the sprayed rollers with an undercoating and of those without an undercoating. This would be due to the fact that the values of shear stresses near the interface between the sprayed layer and the matrix were not always great. The thickness of the sprayed layer and the necessity of undercoating should be designed, taking the depth where the maximum value of shear stress occurs into consideration. 5. Conclusions The rolling contact fatigue of alumina ceramics sprayed on the rollers was investigated under oil lubricating
The authors wish to express their gratitude to Mr. Kazuhiko Hagiwara for his helpful assistance. References [1] Ahmen R, Hadfield M. Rolling contact fatigue performance of plasma sprayed coatings. Wear 1998;220(1):80–91. [2] Nieminen R, Vuoristo P, Niemi K, Ma¨ntyla¨ T, Barbezat G. Rolling contact fatigue failure mechanisms in plasma and HVOF sprayed WCCo coatings. Wear 1997;212(1):66–77. [3] Yoshida M, Tani K, Nakahira A, Nakajima A, Mawatari T. Durability and tribological properties of thermally sprayed WC cermet coating in rolling/sliding contact. Proceedings of ITSC, 1995, Kobe. 1995. p. 663–668. [4] Yoshida A, Fujii M, Nagamori K, Haishi H, Uchimoto H. Study on rolling contact fatigue of fine ceramics (part 4: in a case of pure rolling contact against steel). Jpn J Tribol 1992;37(1):63–74. [5] Doerner MF, Nix WD. A method for interpreting the data from depth-sensing indentation instruments. J Mater Res 1986;1(4):601–9. [6] Oliver WC, Pharr GM. An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments. J Mater Res 1992;7(6):1564–83. [7] Fujita K, Yoshida A. Surface durability of steel rollers (in the case of case-hardened and of nitrided rollers). Bull JSME 1978;21(154):761–7. [8] Smith JO, Liu CK. Stresses due to tangential and normal loads on an elastic solid with application to some contact stress problems. Trans. ASME, J Appl Mech 1953;20:157–66.