Influence of contact stress on rolling contact fatigue of composite ceramic coatings plasma sprayed on a steel roller

Tribology International 73 (2014) 47–56

Contents lists available at ScienceDirect

Tribology International journal homepage: www.elsevier.com/locate/triboint

Influence of contact stress on rolling contact fatigue of composite ceramic coatings plasma sprayed on a steel roller Jia-jie Kang a, Bin-shi Xu b, Hai-dou Wang a,b,n, Cheng-biao Wang a a b

School of Engineering and Technology, China University of Geosciences, Beijing 100083, China National Key Laboratory for Remanufacturing, Academy of Armored Forces Engineering, Beijing 100072, China

art ic l e i nf o

a b s t r a c t

Article history: Received 5 October 2013 Received in revised form 12 December 2013 Accepted 23 December 2013 Available online 15 January 2014

The influence of contact stress on rolling contact fatigue performance of plasma sprayed Al2O3–40 wt% TiO2 composite ceramic coating was investigated using a double-roll test machine under pure rolling contact condition. The shear stresses within the coating were analyzed with the three-dimensional finite element method. Three modes of failures, i.e., surface abrasion, spalling, and delamination, were observed during this investigation. The failure mechanisms of surface abrasion, spalling, and delamination were discussed in detail. The initiation and propagation of fatigue cracks were mainly caused by the shear stress, which were highly influenced by the contact stress. & 2014 Elsevier Ltd. All rights reserved.

Keywords: Rolling contact fatigue Thermal spray coatings Failure mechanism

1. Introduction Ceramic materials have the superior properties of hightemperature resistance, corrosion resistance, wear resistance, high strength, high hardness, and electric insulation [1,2]. Plasma spraying technology with high flame temperature and fast velocity of particles is very suitable for preparing ceramic coatings with high melting point [3–5]. The preparation of ceramic coatings on a metal substrate can combine the high toughness, processability, and thermal conductivity of metal materials with excellent hightemperature resistance, corrosion resistance and wear resistance of ceramic materials to exert the advantages of the two kinds of materials. This can significantly improve the surface properties of mechanical components which are prone to abrasive wear and corrosion wear. Al2O3 coating is one of the most widely used ceramic coatings in an industrial equipment. However, Al2O3 coatings have the disadvantages of high brittleness and low bonding strength with the metal substrate. Therefore, there are few applications of Al2O3 coating to rolling contact machine elements except for some rolls under relatively low contact stress. When a certain amount of TiO2 is dissolved in Al2O3, the toughness, impact resistance of composite ceramic coatings, and the bonding strength between the coatings and the substrate can be improved. Al2O3–40 wt% TiO2 composite ceramic coatings (AT40

n Corresponding author at: School of Engineering and Technology, China University of Geosciences, Beijing 100083, China. Tel.: þ86 10 66718475; fax: þ86 10 66717144. E-mail address: [email protected] (H.-d. Wang).

0301-679X/$ - see front matter & 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.triboint.2013.12.019

coatings) with high content of TiO2 exhibit relatively high density, toughness, and bonding strength with the substrate. This can greatly improve the rolling contact fatigue (RCF) strength of AT40 coatings. Many studies on the RCF behavior of thermal spray coatings in rolling contacts have indicated that the performance is dependent on the contact stress during the tests [6,7]. The shear stress within the coating due to the contact stress will remarkably influence the RCF failure mechanism [8,9]. There is an increased demand for RCF behavior, reliability, and load bearing capacity of composite ceramic coating and future applications call for their use in more hostile environments. However, there is few research on the RCF failure mechanism under the effect of contact stress of composite ceramic coatings. In the present study, the RCF behavior and failure mechanism of AT40 coatings under different contact stresses were investigated in detail. As well as an RCF failure model was established based on the simulation of shear stresses within the coating and the analysis of the failed coating specimens. 2. Experimental test procedure 2.1. Coating deposition A supersonic plasma spray system was used to deposit AT40 composite ceramic coatings and Ni/Al bonding coatings on the surface of a tempered AISI 1045 steel roller with the hardness of 260 HV0.1. The micro-morphologies of Ni/Al powders and AT40 composite ceramic powders are shown in Fig. 1. It can be seen that the Ni/Al powders and AT40 powders present sphericity and

48

J.-j. Kang et al. / Tribology International 73 (2014) 47–56

Fig. 2. Cross section morphology of the AT40 coating.

2.2. Experimental conditions and procedures

Fig. 1. Micro-morphologies of the spray powders. (a) Ni/Al spray powders and (b) AT40 spray powders.

Table 1 Plasma spraying parameters. Plasma spray parameters

Spraying materials Ni/Al

Argon gas flow (m3/h) Hydrogen gas flow (m3/h) Nitrogen gas flow (m3/h) Spraying current (A) Spraying voltage (V) Spraying distance (mm) Powder feed rate (g/min)

3.4 0.3 0.6 320 140 150 30

AT40 2.8 0.4 0.6 440 110 100 30

irregular polyhedron, respectively. The plasma spray parameters of depositing the composite ceramic coatings and Ni/Al bonding coatings are presented in Table 1. Prior to the plasma spraying process, the surface of substrate roller was sandblasted and preheated to the temperature of about 200 1C. The average thickness of as-sprayed coating on the test roller was 400 μm. Fig. 2 shows the cross section morphology of the AT40 coating, which presents a relative dense micro-structure with a few microdefects.

The rolling contact fatigue (RCF) performance under different contact stress levels was investigated by using a double-roll RCF test machine [10]. The schematic of the test roller and standard roller is shown in Fig. 3(a). The load was applied to the upper roller (test roller) by the hydraulic-lever system. The acoustic emission probe was fixed on the shaft bed to monitor the failure process of the test roller for its sensitivity to plastic deformation and brittle fracture of materials. Only the AE signals over threshold value of 60 dB were collected, as well as the failure point would be judged once the AE count exceeded 350; meanwhile the test machine would stop automatically to keep the original failure morphologies for analyzing. Fig. 3(b) shows the configuration of the test roller and standard roller. AT40 coating was prepared on the external circle surface of the test roller. The length of contact line is 5 mm. The test rollers were ground to give an average coating thickness and surface roughness of 400715 μm and 0.6270.01 μm, respectively. The material of the standard roller was tempered AISI 52100 steel with the hardness of 770 HV0.1. The standard rollers were ground to reach the surface roughness of 0.3670.01 μm. The surface roughness of the test roller and standard roller was measured by an Olympus OLS4000 laser 3D microscope. The micro-hardness, elastic modulus, Poisson’s ratio, and surface roughness of the test roller and standard roller are shown in Table 2. The RCF tests for AT40 coatings were performed under four different contact stresses. The Hertz equation (Eq. (1)) was used to calculate the corresponding loads of four different contact stresses, as presented in Table 3. At least 10 tests were conducted at each contact stress vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi   u F ∑ρ u ð1Þ smax ¼ t 1  υ2 1  υ2  π L E 1 1 þ E2 2 where smax is the maximum contact stress, F is the load, L is the length of the contact line, υ1 and E1 are Poisson’s ratio and elastic modulus of the AT40 coating, respectively, υ2 and E2 are Poisson’s ratio and elastic modulus of the standard roller, respectively, Σρ is the sum of principal curvatures of the test roller and standard roller, which can be calculated by ∑ρ ¼

1 1 1 1 þ þ þ R11 R12 R21 R22

ð2Þ

J.-j. Kang et al. / Tribology International 73 (2014) 47–56

49

Table 3 Corresponding loads of four contact stresses. Load (N)

1247

2217

3464

4988

Contact stress (GPa)

0.75

1

1.25

1.5

Fig. 4. Distribution of maximum shear stress tmax perpendicular to the contact line for the AT40 coating under different contact stresses.

where R11 and R12 are the principal curvatures vertical to and along the rolling direction of the test roller, respectively, and R21 and R22 are the principal curvatures vertical to and along the rolling direction of the standard roller, respectively.

3. Results and discussion 3.1. Shear stresses distribution

Fig. 3. Schematic of the rollers. (a) Assembly of rollers and (b) configuration of the rollers.

Table 2 Mechanical properties of the test roller and standard roller. Roller type

Test roller

Standard roller

Roller materials Microhardness (HV0.1) Elastic modulus (GPa) Poisson’s ratio Surface roughness Ra (μm)

AT40 coating 910 173 0.3 0.62

AISI 52100 steel 770 219 0.3 0.36

In general, the subsurface shear stress induced by contact stress is the dominant factor which causes spalling and delamination failures under pure rolling contact condition. The distribution of maximum shear stress (MSS) and orthogonal shear stress (OSS) within the coating under different contact stresses was analyzed using the three-dimensional finite element method (FEM). Fig. 4 shows the distribution of MSS perpendicular to the contact line within the coating. With the contact stress increasing, the peak value of the MSS increases, and the position of peak value moves away from the coating surface. The MSS decreases at the interface, but still sustains a high level. Fig. 5 shows the distribution of MSS along the contact line within the AT40 coating. When the contact stress S1 ¼0.75 GPa, the MSS of the coating edge is small about 500 MPa, but increases sharply along the coating interior and achieves the peak value (808 MPa) in the 200 μm distance from the edge. Then it gradually decreases and reaches 500 MPa in the 500 μm distance from the edge, and almost keeps constant. Under other contact stresses, S2 ¼1 GPa, S3 ¼1.25 GPa and S4 ¼1.5 GPa, the MSS exhibits similar distribution in the width direction. With increase in contact stress, the peak value of MSS gradually increases, and the distance from the interface continually decreases. The distribution of OSS perpendicular to the contact line within the coating has the similar tendency with that of MSS, as shown in Fig. 6. With the contact stress increasing, the peak value of the OSS increases. Meanwhile, the position of the peak value is located within the coating, and continually moves away from the coating surface. Table 4 shows characteristic parameters of MSS and OSS. Under different contact stress levels, the peak values and the values at

50

J.-j. Kang et al. / Tribology International 73 (2014) 47–56

the interface of MSS are all larger than those of OSS. In addition, the distance of the peak value position from the interface of MSS is smaller than that of OSS.

3.2. RCF failure mechanism Table 5 shows the RCF lives and failure modes of AT40 coating under different contact stresses. The main failure modes under S1 ¼0.75 GPa were spalling and surface abrasion. However, the probability of delamination increased with increasing contact stress.

Fig. 5. Distribution of maximum shear stress tmax along the contact line for the AT40 coating under different contact stresses.

Under the relatively high contact stress levels of S3 ¼ 1.25 GPa and S4 ¼ 1.5 GPa, delamination became the main failure mode. 3.2.1. Surface abrasion Surface abrasion is one of the typical failure modes of thermal sprayed coating, which was defined as the failure caused by micropitting on the surface of wear track [11]. The typical surface abrasion morphology of D9 sample (S4 ¼1.5 GPa; N ¼0.89  104) is shown in Fig. 7. It can be seen from Fig. 7(a) that the surface abrasion exhibits material removal of superficial layer, which was caused by the connection of the micropittings. The surface morphology and three-dimensional profile of the micropitting were observed by the Olympus OLS4000 laser 3D microscope, as shown in Fig. 7(b) and Fig. 7(c), respectively. It is obvious that the diameter and depth of the micropitting are about 90 μm and 30 μm, respectively. By observing the morphologies of the micropittings, it was found that the diameter and depth ranges of the micropittings were 70–100 μm and 20–40 μm, respectively. Surface abrasion failure of plasma sprayed coatings can be associated with asperity contact and microslip within the contact region. Therefore, the lubricant condition has a non-negligible effect on the occurrence of micropitting. The λ ratio, the ratio of lubricant film thickness to the root mean square roughness of the two surfaces, is a common characteristic parameter to characterize the lubricant condition between the test roller and standard roller. When λ 43, the rollers are in a fluid lubricant state. The friction surface will be completely separated by the continuous oil film which bears the load to decrease abrasion. When λ o 1, the boundary lubricant film generated by the effect of oil and friction surfaces, coupled with asperity, will bear the load. When 1o λ o3, the rollers are in a mixed lubricant state. The load will be jointly born by the fluid lubricant film, boundary lubricant film and Table 5 The RCF lives and failure modes of AT40 coating under different contact stresses. No A: S1 ¼ 0.75 GPa

1 2 3 4 5 6 7 8 9 10 a

Fig. 6. Distribution of orthogonal shear stress txy perpendicular to the contact line for the AT40 coating under different contact stresses.

b c

B: S2 ¼ 1 GPa

C: S3 ¼ 1.25 GPa

D: S4 ¼ 1.5 GPa

N (  104 cycles)

Failure mode

N (  104 cycles)

Failure mode

N (  104 cycles)

Failure mode

N (  104 cycles)

Failure mode

3.12 3.34 3.75 3.99 4.37 4.48 4.65 4.92 5.13 5.3

DEa SPb SP SAc SP SP SA SP SA SP

1.45 1.62 1.97 2.19 2.37 2.5 2.91 3.1 3.49 3.71

DE DE SP DE SP SA SA SP SP SA

0.47 0.61 0.72 0.79 0.85 1.03 1.12 1.2 1.46 1.74

DE DE DE DE DE SA DE SA SP SA

0.41 0.5 0.62 0.64 0.66 0.72 0.78 0.79 0.89 1.09

DE DE DE DE DE DE DE DE SA SA

Delamination. Spalling. Surface abrasion.

Table 4 Characteristic parameters of maximum shear stress tmax and orthogonal shear stress txy. Stress type

Characteristic parameter

Contact stress (GPa) 0.75

1

1.25

1.5

Maximum shear stress tmax

Peak value (MPa) The distance of peak value position from the interface (μm) The value at the interface (MPa)

808 200 634

1335 175 1179

1780 148 1642

2243 120 2112

Orthogonal shear stress txy

Peak value (MPa) The distance of peak value position from the interface (μm) The value at the interface (MPa)

365 228 268

647 204 514

965 176 833

1268 147 1174

J.-j. Kang et al. / Tribology International 73 (2014) 47–56

51

Fig. 7. Typical surface abrasion morphology of D9 sample (S4 ¼ 1.5 GPa; N ¼ 0.89  104). (a) Morphology of surface abrasion, (b) morphology of micropitting and (c) threedimensional profile of micropitting.

asperity. The oil film parameter h

m ffi λ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

R2a1 þ R2a2

λ can be calculated by [12] ð3Þ

where hm is the minimum lubricant film thickness, Ra1 is the root mean square average roughness of the surface of the test roller (0.62 μm), and Ra2 is the root mean square average roughness of the surface of the standard roller (0.36 μm).

The elastohydrodynamic lubrication theory (EHL) was used to analyze the lubrication problem of the friction pair. The formula of minimum film thickness can be expressed as hm ¼ 2:65α0:54 ðη0 uÞ0:7 R0:43 E0  0:03 ðL=WÞ0:13

ð4Þ

where α is viscosity–pressure coefficient (0.038 mm /N), η0 is fluid viscosity at atmospheric pressure (0.581 Ns/m2), u is mean velocity (0.628 m/s), R is equivalent radius of curvature (0.015 m), E0 is equivalent elastic modulus, which can be calculated by Eq. (5), L is 2

52

J.-j. Kang et al. / Tribology International 73 (2014) 47–56

Table 6 The surface roughness of the rollers in different RCF stages under different contact stresses. Contact stress

Material

Cycles 0 (Original state, μm)

102 (μm)

103 (μm)

104 (μm)

S1 ¼ 0.75 GPa

AT40 coating AISI 52100 steel

0.62 7 0.01 0.36 7 0.01

0.6770.03 0.28 7 0.02

0.81 7 0.03 0.20 7 0.02

0.93 70.03 0.187 0.02

S2 ¼ 1 GPa

AT40 coating AISI 52100 steel

0.62 7 0.01 0.36 7 0.01

0.69 7 0.03 0.29 7 0.02

0.82 7 0.03 0.22 7 0.02

1.02 7 0.03 0.19 70.02

S3 ¼ 1.25 GPa

AT40 coating AISI 52100 steel

0.62 7 0.01 0.36 7 0.01

0.75 7 0.05 0.317 0.03

0.84 7 0.05 0.277 0.03

1.137 0.05 0.22 70.03

S4 ¼ 1.5 GPa

AT40 coating AISI 52100 steel

0.62 7 0.01 0.36 7 0.01

0.78 7 0.06 0.3570.04

0.89 7 0.05 0.30 7 0.04

1.26 7 0.04 0.26 70.04

will cause abrasive wear and adhesive wear to induce the surface abrasion failure [13].

Fig. 8. The variation of surface roughness of AT40 coating and AISI 52100 steel at different number of cycles under different contact stresses.

the length of the contact line (0.05 m), and W is the unit load along contact length. The equivalent elastic modulus E0 can be calculated by E0 ¼

2E1 E2 ð1  υ21 ÞE1 þ ð1  υ22 ÞE2

ð5Þ

where E1 and E2 are the elastic moduli of AT40 coating and AISI 52100 steel, respectively, and υ1 and υ2 are Poisson’s ratios of AT40 coating and AISI 52100 steel, respectively. The calculated value of equivalent elastic modulus E0 is 212.42 GPa. The surface roughness of the rollers in different RCF stages under different contact stresses was measured to investigate the variation of lubrication state between the rollers, as shown in Table 6 and Fig. 8. It can be seen from Fig. 8 that the surface roughness of AT40 coating gradually increases with increase in number of cycles. This is mainly due to fracture and adhesion of the coating under the effect of abrasive wear and adhesive wear. The generated worn debris will aggravate the wear to increase the surface roughness. However, the surface roughness of AISI 52100 steel shows the opposite variation of that of the AT40 coating. This is attributed to the excellent toughness and machinability of AISI 52100 steel. Therefore, under the polishing effect of micro-debris in lubricant oil, the surface roughness of AISI 52100 steel significantly decreases. The minimum lubricant film thickness hm and λ ratios at different number of cycles under different contact stresses calculated by Eq. (3) and Eq. (4) are presented in Table 7. With increase in contact stress, the friction pair tends to enter the boundary lubrication state as early as possible. At this moment, the asperity on the surfaces of AT40 coating and AISI 52100 steel will directly contact with each other. The accumulation of the AT40 coating debris which is generated by plastic formation and micro-facture

3.2.2. Spalling Fig. 9 shows the typical spalling failure morphology of A5 sample (S1 ¼0.75 GPa; N¼ 4.37  104). The spalling pit presents a circular shape with smooth leading edge and an unmelted particle, as shown in Fig. 9(a). The bottom near the trailing edge shows beach strip shape which is a typical morphology of fatigue crack propagation, and there are a number of annular cracks around the trailing edge (Fig. 9(b)). The diameter and depth of the spalling pit of A5 sample are 760 μm and 100 μm, respectively (Fig. 9(c)). By observing the morphologies of the spalling pits, it was found that the diameter and depth of the spalling pits were in the ranges of 600–800 μm and 80– 110 μm, respectively. The failure mechanism of spalling is complicated. Hogmark and Hedenqvist [14] considered that spalling failure is related to the surface wear behavior and the microstructure of coatings, and spalling origins from the surface or subsurface of coatings. Furthermore, Zhang et al. [15] considered that the generation mechanism of spalling is the material removal of superficial layer, which is caused by the initiation and propagation of cracks on the subsurface of coatings. The FEM results indicated that the depths of peak values of MSS and OSS were much larger than those of the spalling pits. Therefore, the main inducement of spalling was the microstress within the coating caused by contact stress rather than MSS or OSS, which was in agreement with the analysis by Zhang et al. [16]. For the relatively high porosity and low cohesion strength of AT40 coating, when AT40 coating was under alternating contact stresses in the RCF process, the microcracks tend to initiate from the micro-defects and unmelted particles, and propagate randomly. When the cracks connect with each other, the main crack as shown in Fig. 10 will be generated. The unstability and fracture of the coating materials near the main crack are the main reasons of spalling failure. Once the annular cracks at the trailing edge of the spalling pit propagated and connected with the secondary crack on the subsurface, further fracture of the coating will happen to expand the spalling pit. 3.2.3. Delamination Delamination is the main failure mode of AT40 coating under high contact stresses (S3 ¼ 1.25 GPa; S4 ¼ 1.5 GPa). Figs. 11 and 12 show the typical surface morphologies of delamination failure of C3 sample (S3 ¼ 1.25 GPa; N ¼0.72  104) and D5 sample (S4 ¼ 1.5 GPa; N ¼ 0.66  104), respectively. The common characteristics of delamination failure are as follows: (1) the area of delamination is large, about 30–100% of coating width; (2) the metal matrix was exposed at the delamination zone, indicating that the coating was separated with the substrate at the interface; (3) the exposed metal presented rough morphology after sand blasting, without any

J.-j. Kang et al. / Tribology International 73 (2014) 47–56

53

Table 7 The minimum lubricant film thickness hm, λ ratios, and lubrication states at different number of cycles under different contact stresses. Contact stress

S1 ¼ 0.75 GPa

S2 ¼ 1 GPa

S3 ¼ 1.25 GPa

S4 ¼ 1.5 GPa

Characteristic parameter

Cycles 0 (Original state)

102

103

104

hm(μm) λ ratio Lubrication state

0.964 1.344 Mixed lubrication

1.328 Mixed lubrication

1.156 Mixed lubrication

1.018 Mixed lubrication

hm(μm) λ ratio Lubrication state

0.894 1.247 Mixed lubrication

1.195 Mixed lubrication

1.053 Mixed lubrication

0.862 Boundary lubrication

hm(μm) λ ratio Lubrication state

0.844 1.177 Mixed lubrication

1.040 Mixed lubrication

0.957 Boundary lubrication

0.733 Boundary lubrication

hm(μm) λ ratio Lubrication state

0.805 1.123 Mixed lubrication

0.942 Boundary lubrication

0.857 Boundary lubrication

0.626 Boundary lubrication

Fig. 9. Typical spalling failure morphology of A5 sample (S1 ¼0.75 GPa; N ¼4.37  104). (a) Morphology of spalling pit, (b) morphology of trailing edge and (c) threedimensional profile of spalling pit.

54

J.-j. Kang et al. / Tribology International 73 (2014) 47–56

Fig. 13. Morphology of interfacial crack of C2 sample (S3 ¼ 1.25 GPa; N¼ 0.61  104).

Fig. 10. Typical cross-section morphology of spalling failure of A10 sample (S1 ¼ 0.75 GPa; N ¼ 5.3  104).

Fig. 11. Typical delaminaiton failure morphology of C3 sample (S3 ¼ 1.25 GPa; N ¼1.25  104).

Fig. 12. Typical delaminaiton failure morphology of D5 sample (S4 ¼ 1.5 GPa; N ¼0.66  104).

obvious worn characteristics such as plowing or adhesion (Fig. 12), showing that delamination was an abrupt unstability failure process after a certain degree of damage accumulation; and (4) the edges of delamination were cliffy with a few of annular cracks. Previous studies related to RCF failure modes of plasma sprayed coatings indicated two distinct modes of delamination failure, i.e. cohesive delamination and adhesive delamination. The term cohesive delamination was defined as the delamination within the coating microstructure, while adhesive delamination was defined as the delamination at the interface between the coating and the substrate [11]. In the present study, all of the delamination failure samples exhibited adhesive delamination. MSS and OSS are key factors of delamination failure of plasma sprayed coatings. However, there is no recognized conclusion that

Fig. 14. Surface and cross section morphologies of longitudinal cracks of C7 sample (S3 ¼ 1.25 GPa; N ¼1.12  104). (a) Surface morphology of longitudinal cracks and (b) cross section morphology of vertical cracks.

which one has the largest effect on the behavior of initiation and propagation of the fatigue cracks within the coatings. The study by Ahmed and Hadfield indicated that the fatigue cracks tend to initiate and propagate at the depth of peak values of MSS and OSS because of the stress concentrations due to the defects in the coating microstructure under the contact stress [12]. The connection of the cracks caused by MSS and OSS and the propagation to the surface of the coatings will induce the removal of coating materials to cause the cohesive delamination or adhesive delamination failures. Many scholars considered that the adhesive

J.-j. Kang et al. / Tribology International 73 (2014) 47–56

55

Fig. 15. Process of delamination failure of AT40 coating.

delamination was mainly due to that the fact the peak value of MSS is large and deep form the coating surface [17]. Fig. 13 shows the morphology of interfacial crack of C2 sample (S3 ¼1.25 GPa; N ¼ 0.61  104). The fatigue crack propagated along the interface between the coating and the substrate. Under the contact stress of 1.25 GPa, the MSS at the interface was 1642 MPa, about twice as large as OSS (833 MPa). Therefore, it was considered that MSS was the key factor of initiation and propagation of interfacial cracks. In the case of supersonic plasma spraying coatings, the high velocity and temperature improve the microstructure and properties of coatings, and thus improve the cohesive strength. However, the contamination before spraying, wettability of impacting lamella, shadowing effect, quenching stress and thermal mismatch stress, and elastic mismatch at the coatings–substrate interface play an important part in affecting the adhesive strength between the coating and the substrate [11]. There is thus inevitably a higher tendency of microdefects at the coating–substrate interface, which results in adhesive delamination under the main driving force of MSS [18]. Fig. 14 shows the surface and cross section morphologies of longitudinal cracks of C7 sample (S3 ¼ 1.25 GPa; N ¼1.12  104). Fig. 14(b) shows the cross section morphology of the longitudinal cracks (slit along the dash line shown in Fig. 14(a)). There were three vertical cracks through the coating. In the direction of coating width, the peak value of MSS was located near the coating edge (shown in Fig. 5), where the fatigue cracks within the coating were inclined to be induced and propagated randomly. Due to this, coupled with the effect of contact stress, the fatigue cracks propagated along the vertical direction and connected with each other, to cause the adhesive delamination failure of the coating. To sum up, the generation mechanism of delamination for AT40 coating can be described as follows: interfacial cracks first initiated from the interface between the coating and the substrate at the position of peak value of MSS; meanwhile, the fatigue microcracks nucleated at the microdefects within the coating, and then extended along the vertical direction and connected with each other under the effects of shear stress and surface pressing force; when the vertical cracks extended to the interface and connected with the interfacial cracks, edge delamination occurred; with continuous initiation and propagation of the vertical cracks, further delamination failure would occur. Fig. 15 shows the process of delamination failure of AT40 coating.

4. Conclusions The RCF performance of plasma sprayed Al2O3–40 wt% TiO2 composite ceramic coatings under four different contact stresses (0.75 GPa, 1 GPa, 1.25 GPa, and 1.5 GPa) was investigated by using a double-roll RCF test machine. The main RCF failure modes of the AT40 coatings were surface abrasion, spalling, and delamination. The probability of delamination increased with increasing contact stress.

The distribution of MSS and OSS within the coating under different contact stresses was analyzed using the FEM. The results showed that the peak values of the MSS and OSS increased with increase in contact stress, and the peak values and the values at the interface of MSS were all larger than those of OSS; the depth of peak value position of MSS is larger than that of OSS. The failure mechanisms of surface abrasion, spalling, and delamination were discussed. (1) Under the boundary lubrication state, the contact of the asperity on the surfaces of friction pairs and the accumulation of the debris in the RCF process will cause the surface abrasion failure. (2) The microcracks tend to initiate from the micro-defects and unmelted particles, and propagate randomly under alternating contact stresses in the RCF process. The main crack will be generated when the cracks connect with each other. The unstability and fracture of the coating materials near the main crack are the main reasons of spalling failure. (3) Interfacial cracks first initiate from the interface under the effect of MSS; meanwhile, the fatigue microcracks nucleate at the microdefects, and then extend along the vertical direction and connect with each other under the effects of shear stress and surface pressing force; when the vertical cracks extend to the interface and connect with the interfacial cracks, delamination occurs. In other words, the RCF failure behavior of AT40 coating is closely related to the micro-defects within the coating, the bonding strength between the coating and the substrate, the asperity contact between the frictional pair, and the distribution of shear stress. The shear stress increases sharply under high contact stress, and tends to induce the initiation and propagation of fatigue cracks within the coating and at the interface, which is the main cause of spalling and delamination failures.

Acknowledgments The paper was financially supported by Distinguished Young Scholars of NSFC (51125023), 973 Project (2011CB013405), NSFC (51275151), NSF of Beijing (3120001), and Fundamental Research Funds for Central Universities.

References [1] Vite J, Vite M, Castillo M, Laguna-Camacho JR, Soto J, Susarrey O. Erosive wear on ceramic materials obtained from solid residuals and volcanic ashes. Tribol Int 2010;43:1943–50. [2] Lu LH, Zhang JW, Shen DJ, Wu LL, Jiang GR, Li L. TEM analysis and wear resistance of the ceramic coatings on Q235 steel prepared by hybrid method of hot-dipping aluminum and plasma electrolytic oxidation. J Alloys Compd 2012;512:57–62. [3] JB Cheng, XB Liang, BS Xu, YX Wu. Characterization of mechanical properties of FeCrBSiMnNbY metallic glass coatings. J Mater Sci 2009;44:3356–63. [4] Cheng JB, Liang XB, Xu BS, Wu YX. Formation and properties Fe-based amorphous/nanocrystalline coating prepared by wire arc spraying process. J Non-Cryst Solids 2009;355:1673–8. [5] Cheng JB, Liang XB, Xu BS, Wu YX. Microstructure and mechanical characteristics of iron-based coating prepared by plasma transferred arc cladding process. Mater Sci Eng A 2008;492:407–12.

56

J.-j. Kang et al. / Tribology International 73 (2014) 47–56

[6] Piao ZY, Xu BS, Wang HD, Wen DH. Investigation of acoustic emission source of Fe-based sprayed coating under rolling contact. Int J Fatigue 2013;4:184–8. [7] Piao ZY, Xu BS, Wang HD. Investigation of spalling mechanism of the thermal sprayed coating under rolling contact by FIB-SEM. Eng Failure Anal 2012;25: 106–11. [8] Piao ZY, Xu BS, Wang HD, Wen DH. Influence of surface nitriding treatment on rolling contact behavior of Fe-based plasma sprayed coating. Appl Surf Sci 2013;1:420–5. [9] Piao ZY, Xu BS, Wang HD, Wen DH. Influence of surface roughness on rolling contact fatigue behavior Fe–Cr alloy coatings. J Mater Eng Perform 2013;2:767–73. [10] Kang JJ, Xu BS, Wang HD, Wang CB. Investigation of a novel rolling contact fatigue/wear competitive life test machine faced to surface coating. Tribol Int 2013;66:249–58. [11] Ahmed R. Contact fatigue failure modes of HVOF coatings. Wear 2002;253: 473–87. [12] Ahmed R, Hadfield M. Failure modes of plasma sprayed WC–15%Co coated rolling elements. Wear 1999;230:39–55.

[13] Holmberg K, Matthews A, Ronkainen H. Coatings tribology-contact mechanisms and surface design. Tribol Int 1998;31:107–20. [14] Hogmark S, Hedenqvist P. Tribological characterization of thin hard coatings. Wear 1994;179:147–54. [15] Zhang XC, Xu BS, Xuan FZ, Tu ST, Wang HD, Wu YX. Rolling contact fatigue behavior of plasma-sprayed CrC–NiCr cermet coatings. Wear 2008;265: 1875–83. [16] Zhang XC, Xu BS, Xuan FZ, Tu ST, Wang HD, Wu YX. Fatigue resistance of plasma-sprayed CrC–NiCr cermet coatings in rolling contact. Appl Surf Sci 2008;254:3734–44. [17] Zhang ZQ, Li GL, Wang HD, Xu BS, Piao ZY, Zhu LN. Investigation of rolling contact fatigue damage process of the coating by acoustics emission and vibration signals. Tribol Int 2012;47:25–31. [18] Li GL, Zhang ZQ, Wang HD, Xu BS, Piao ZY, Zhu LN. Acoustic emission monitoring and failure mechanism analysis of rolling contact fatigue for Febased alloy coating. Tribol Int 2013;61:129–37.