- Email: [email protected]
Thermomechanical fatigue behavior of an air plasma sprayed thermal barrier coating system
Materials Science and Engineering A 528 (2011) 8396–8401
Contents lists available at SciVerse ScienceDirect
Materials Science and Engineering A journal homepage: www.elsevier.com/locate/msea
Thermomechanical fatigue behavior of an air plasma sprayed thermal barrier coating system Z.B. Chen a,c , Z.G. Wang a,∗ , S.J. Zhu b a b c
Shenyang National Laboratory for Materials Science, Institute of Metal Research, Chinese Academy of Sciences, 72 Wenhua Road, Shenyang 110016, PR China Department of Intelligent Mechanical Engineering, Fukuoka Institute of Technology, Higashi-ku, Fukuoka 811-0295, Japan Shenyang Aeroengine Research Institute, Aviation Industry Corporation of China, 1 Wanlian Road, Shenyang 110015, PR China
a r t i c l e
i n f o
Article history: Received 7 December 2010 Received in revised form 22 June 2011 Accepted 19 August 2011 Available online 25 August 2011 Keywords: Thermal barrier coating Thermomechanical fatigue Air plasma spraying
a b s t r a c t Failure behavior of an air plasma sprayed thermal barrier coating (TBC) system was investigated under inphase (IP) and out-of-phase (OP) thermomechanical fatigue (TMF) tests. All the TMF tests were performed in the temperature range of 450–850 ◦ C with a given period of 300 s under mechanical strain control. Both the bond coat NiCrAlY and the top coat 7%Y2 O3 –ZrO2 were fabricated by air plasma spraying (APS). Results revealed that the IP TMF lifetime was longer than that of the OP TMF under the same mechanical strain amplitude. Morphology observations of the failed specimens showed that the coating cracking and spallation processes were different in the two phase conditions. © 2011 Elsevier B.V. All rights reserved.
1. Introduction Thermal barrier coating (TBC) systems have been widely used to hot section parts of gas turbine engines to provide thermal protection to metallic components [1–3]. They offer potential for lowering the temperature of the metal substrate and thus produce an increase in power efficiency and a decrease in greenhouse gas emission. A typical TBC system consists of a superalloy substrate, a bond coat and a ceramic top coat, usually ZrO2 stabilized with 6–8 wt% Y2 O3 [4]. In service, hot section components experience severe cyclic temperature gradients and mechanical loads. As a consequence, thermomechanical fatigue (TMF), which provides a closer simulation of the actual strain-temperature cycle in an engine environment, is a major life limiting factor for gas turbine blades [5]. Therefore, TMF tests are useful for evaluating the service lifetime of a TBC system, for identifying the damage mechanism, and for a life modeling approach. However, only a few publications are available in the open literature with regard to TMF tests of TBC systems [6–15]. The main reasons for the limited number of publications are: (1) the difficulty of obtaining sufficiently powerful heating and cooling technique associated with the poor heat absorption and conductivity of the ZrO2 top coat in the TBC system [9,11], and (2) establishing a service-like temperature gradient across the thickness of the TBC system.
∗ Corresponding author. Tel.: +86 24 8397 8870; fax: +86 24 2389 1320. E-mail address: [email protected] (Z.G. Wang). 0921-5093/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2011.08.031
Different heating equipments have been adopted in TMF tests for TBC systems, such as radiation furnaces [6,12], direct or indirect induction heating [8–11], and lamp furnaces [7,13–15]. The first published paper reported the results for a silicon carbide igniter furnace for heating a TBC system, which produced a temperature gradient around 50 ◦ C during transient heating and less than 20 ◦ C during a high temperature dwell [6]. Then induction heating was used to perform a TMF test on a TBC system and obtained an inverse temperature gradient across the coating that was contrary to the in-service situation [8]. Later, an indirect induction heating method was developed and produced a temperature difference of 50 ◦ C between the top coat and substrate during the high temperature hold, but the heat transfer susceptor was hard to select and the test period was long [9–11]. A heating system closer to the ideal was the lamp furnace, which could heat specimens up to 1000 ◦ C in a few seconds [13–15] and obtain a temperature gradient of about 170 ◦ C [14,15]. Among these studies, it was found that the relationship of the lifetime for TBC systems between in-phase (IP) TMF and out-of-phase (OP) TMF is variable. Wright [6] found that the OP TMF lifetime of the TBC system was longer than the IP TMF one, whereas Baufeld et al.’s experiments [10] revealed that the IP TMF lifetime was longer. The cracking and failure mechanisms of TBC systems reported by different authors were also variable. Although TBC systems typically fail by coating delamination or spallation under TMF conditions [6,8–11,13,15], the crack initiation sites are different. Cracks may develop from the uncoated inner surface [10,11], the bond coat [8,10], or the interface between the bond coat and the top coat [9]. As a result, it is difficult to understand the fracture behavior of TBC systems under service conditions and to predict the lifetime of the components.
Z.B. Chen et al. / Materials Science and Engineering A 528 (2011) 8396–8401
8397
Fig. 1. Dimensions of TBC specimens for thermomechanical fatigue tests.
The bond coat, usually MCrAlY overlay coating (where M is Ni, Co, or a combination of these elements), is the most crucial component of the TBC system [16] that can be produced by thermal spraying, sputtering or evaporation; however, for practical applications, thermally sprayed coatings are normally preferred [17]. Among thermal spraying techniques, APS and high velocity oxygen fuel (HVOF) can be conducted at atmospheric pressure, therefore their equipment investment and production costs are low compared with vacuum plasma spraying (VPS). Meanwhile, APS shows a considerably better efficiency than other methods [18]. Thus it is attractive to spray the bond coat by APS. The aim of this paper is to present experimental results concerning TMF lifetime and failure behavior of a TBC system consisting of a Ni-based superalloy substrate, a NiCrA1Y bond coat and a Y2 O3 –ZrO2 top coat, both of which were fabricated by APS. The relationship of the lifetime and differences of the cracking behaviors between the IP and OP conditions are discussed. 2. Experimental The cast Ni-based superalloy M963 was used as a substrate. Its chemical composition (in wt.%) was C 0.15, Cr 8.89, Al 6.00, Ti 2.55, Mo 1.64, W 10.1, Co 10.0, Nb 1.10, Zr 0.03, B 0.03, Ce 0.02, Y 0.01 and Ni balance. Before machining, the specimens were solution-treated at 1210 ◦ C for 4 h followed by air cooling. Cylindrical tube specimens were machined with a total length of 135 mm, an internal diameter of 7 mm and in the gauge length an external diameter of 10 mm as shown in Fig. 1. Before spraying these specimens, all the substrates were grit blasted by alumina powder with 80 mesh grain size distribution. The coating substrate was deposited by APS with a Ni–25Cr–5Al–0.5Y alloy as the bond coat with a normal thickness of 130 m and 7%Y2 O3 –ZrO2 as the top coat with 250 m thickness. The APS setting was METCO 7 M and spray parameters for the bond coat and the top coat are listed in Table 1. The TMF tests were performed on an MTS810 closed-loop servohydraulic testing machine with computer control. A radiation furnace powered by four cylindrical quartz lamps, each with a maximum power of 2.5 kW, was used for heating. The radiation of the quartz lamps was focused on the specimen by reflection from elliptical mirrors. The outer surface temperature of the specimen, which was controlled by the power output of the lamps, was measured with a thermocouple enlacing the specimen. Cooling was obtained by internally compressed air combined with reducing the power output of the lamps. Axial strain amplitudes were controlled using a self-supporting extensometer that had
a gauge length of 23 mm, supported with ceramic rods. A triangle waveform was used for both thermal cycling and mechanical cycling. Two kinds of TMF tests were used: IP where the maximum mechanical strain coincided with the maximum temperature and OP where the maximum mechanical strain was attained at the minimum temperature. TMF tests were carried out in the temperature range of 450–850 ◦ C with a cyclic period of 300 s under mechanical strain control. The mechanical strain amplitudes were varied from 0.35% to 0.5% with a strain ratio of −1. The apparatus yielded a temperature difference of about 90 ◦ C between the top coat and the substrate during dwells at the top coat temperature of 900 ◦ C. After the TMF tests, the cyclically induced morphologies of all specimens, were recorded by a high accuracy digital camera, and fracture surfaces were investigated using a scanning electron microscope (SEM). Longitudinal sections were cut from the tested specimens and were embedded in an epoxy resin. Then these sections were metallographically prepared and observed by the SEM. 3. Results 3.1. Cyclic deformation and fatigue lifetime Fig. 2 shows the typical stress–mechanical strain hysteresis loops for IP TMF and OP TMF at a mechanical strain amplitude of 0.45%. It can be seen that the maximum and minimum stress of each hysteresis loop are unsymmetrical. In IP TMF, the value of the maximum stress is lower than that of the minimum one, while it is opposite under OP TMF, and the value of the maximum stress is higher. The cyclic stress response behavior at a mechanical strain amplitude of 0.45% is shown in Fig. 3, which displays the variation of the maximum stress ( max ) and the minimum stress ( min ) as well as the mean stress ( m ) with the number of cycles (N). In the tensile stress cycles, the maximum stress ( max ) gradually decreases in IP TMF but increases under OP TMF with an increase in N. In the compressive stress cycles, the tendency of the minimum stress ( min )
Table 1 Spray parameters for bond coat and top coat. Bond coat Spraying distance (mm) Feed rate (g/min) Carrier gas (Ar) pressure (MPa) Voltage (V) Current (A)
145–150 30–45 0.40–0.55 40–60 500–530
Top coat 70–120 35–45 0.40–0.55 45–65 500–550
Fig. 2. Middle-life hysteresis loops for IP and OP TMF at εmech /2 = 0.45%.
8398
Z.B. Chen et al. / Materials Science and Engineering A 528 (2011) 8396–8401
The hysteresis loops and cyclic stress response behaviors of the TBC system are similar to the corresponding material of the substrate M963, while the relationship of the fatigue life for the TBC system between the IP and OP conditions is different from that of the substrate [19]. For the bare substrate, IP TMF and OP TMF fatigue lives with regard to the mechanical strain amplitude exhibit a crossover. The IP TMF life is shorter than the OP TMF life at high mechanical strain amplitudes whereas it is longer than the OP TMF life at low mechanical strain amplitudes, which is mainly due to creep and oxidation variations of the specimens. However, creep and oxidation do not influence the lifetime of the superalloy with TBC because the coating prevents the substrate from damage at high temperature in the TBC system; moreover, the highest temperature in this test program was just 850 ◦ C. 3.2. Morphology observation Fig. 3. Cyclic stress response curves for IP and OP TMF at εmech /2 = 0.45%.
Fig. 4. Mechanical strain amplitude as a function of cycles to failure.
is just opposite. As a result, the mean stress ( m ) is compressive in IP TMF but tensile in OP TMF. The fatigue life for IP and OP TMF as a function of mechanical strain amplitude is shown in Fig. 4. It can be seen that the fatigue life decreases with an increase in the mechanical strain amplitude. Under the same mechanical strain amplitude, OP TMF exhibits a shorter fatigue life than IP TMF.
Typical photos of failed specimens recorded by a digital camera are shown in Fig. 5. It can be seen that the failure behaviors of the specimens are different between the two phase conditions. The specimens failed in the gauge lengths in the IP tests for all four mechanical strain amplitudes (Fig. 5(a)); however, under the OP condition, at all the four mechanical strain amplitudes, there was cracking in the coating though the specimens also failed within the gauge lengths (Fig. 5(b)). From the inset of Fig. 5(b) it is shown that the coating spallation was located at the interface of the bond coat/substrate. A representative fractograph of failed specimens under the IP TMF condition is shown in Fig. 6(a). No crack is found in the coating and the interfaces are almost intact. By contrast, the fracture surface of the specimen failed in the OP TMF test shows a different failure behavior in that the bond coat is widely detached from the substrate (Fig. 6(b)). The fracture surfaces are consistent with the appearance of the TMF failed specimens as shown in Fig. 5. Typical longitudinal sections close to a fracture surface are shown in Fig. 7. No obvious delamination can be seen in the IP TMF tests as shown in Fig. 7(a). A crack is detected only in the top coat as indicated by the arrow. However, there is a crack passing through both the top coat and the bond coat under the OP TMF condition (Fig. 7(b)). The crack does not propagate into the substrate but is deflected along the interface between the bond coat and the substrate, causing interface delamination. Cracks can also be found at the interface of the top coat/bond coat. The phenomena observed from the longitudinal sections are in accordance with the macroscopic surface and fracture surface observations.
Fig. 5. Photos of failed specimens at (a) IP TMF and (b) OP TMF, inset of (b) showing spallation of the coating.
Z.B. Chen et al. / Materials Science and Engineering A 528 (2011) 8396–8401
8399
Fig. 6. Fractographs of failed specimens at (a) IP TMF and (b) OP TMF (TC, top coat; BC, bond coat; Sub, substrate).
Fig. 7. SEM images of longitudinal sections of fracture specimens at (a) IP TMF and (b) OP TMF.
neglected. Besides the applied mechanical strain, there is a thermal mismatch strain arising from the thermal expansion mismatch between the coating and substrate during the TMF testing. The thermal expansion of the coating is taken to be that of the bond coat because the damage occurred at the bond coat/substrate interface. The thermal strain εth of the coating or the substrate is calculated as follows: εth = ˛ T
(1)
where ˛ is the mean coefficient of thermal expansion in the temperature range of T. The specimen usually fails at the maximum tensile stress which corresponds to the maximum and minimum temperature for IP TMF or OP TMF, respectively. Therefore, the maximum thermal mismatch strain range εth between the coating and the substrate in the TMF tensile cycle can be calculated as: Fig. 8. Sketch of strains in a coated specimen.
εth = ˛M963 (Tmax − Tmiddle ) − ˛BC (Tmax − Tmiddle )
(2.1)
εth = ˛M963 (Tmin − Tmiddle ) − ˛BC (Tmin − Tmiddle )
(2.2)
4. Discussion According to the experimental results, the TBC system can be considered as a two part composite: the substrate, and the combination of the top coat with the bond coat. A simple model is proposed as shown in Fig. 8 where the coating consists of the bond coat and the top coat. For simplicity, the coating and substrate are assumed to experience the same applied mechanical strain; effects of the interface bonding strength on the cracking behaviors are
Eq. (2.1) applies to the IP TMF test and the maximum thermal mismatch strain range arrives at the maximum temperature 850 ◦ C. Eq. (2.2) applies to the OP TMF test and the maximum thermal mismatch strain range obtains at the minimum temperature 450 ◦ C. Tmiddle is the mid temperature of the test, which is 650 ◦ C. The parameters such as elastic modulus, coefficient of thermal expansion and Poisson ratio are shown in Fig. 9(a)–(c), respectively [20,21].
8400
Z.B. Chen et al. / Materials Science and Engineering A 528 (2011) 8396–8401
Fig. 9. (a) Elastic modulus and (b) coefficient of thermal expansion as a function of temperature for substrate and bond coat; (c) Poisson ratio as a function of temperature for bond coat [20,21].
Fig. 10. Strain analysis of the coating in the axial direction for (a) IP TMF and (b) OP TMF.
4.1. Relationship of fatigue lifetime to IP TMF and OP TMF The total axial strain on the coating, i.e. the sum of the applied mechanical strain and the thermal mismatch strain, is a crucial factor for the coating life since it is directly related to the applied axial stress. During IP TMF cycling, the temperature increases from the mid-range to the maximum, and the thermal mismatch strain of the coating is positive from Eq. (2.1) because the coefficient of thermal expansion of the coating is lower than that of the substrate. The thermal strain adds to the applied mechanical strain, producing a larger total axial strain range as shown in Fig. 10(a). Whereas during OP TMF cycling, the temperature decreases from the mid-range to the minimum, and the thermal mismatch strain of the coating is negative in Eq. (2.2), which is of opposite sign to that of the applied mechanical strain. Therefore the total axial strain obtains a lower strain range (Fig. 10(b)). However, the elastic moduli of the coating differ between the maximum and minimum temperature. When the total axial strain is multiplied by the elastic modulus, the experienced axial stress can be calculated as shown in Table 2. It
Table 2 Axial stress in the coating. Applied mechanical strain (%) 0.50 0.45 0.40 0.35
Axial stress (MPa) IP
OP
716.23 646.85 577.48 508.10
823.71 738.09 652.46 566.84
is found that the stress in the coating under OP TMF is higher than that under IP TMF at the same applied mechanical strain. Therefore, a shorter fatigue lifetime under OP TMF is expected. The differences of fatigue lifetime between IP and OP TMF found here is in conflict with results reported in the open literature. For example, it was reported [6] that OP TMF life was longer than IP TMF life, which was believed to be governed by the extent of compression at minimum cycle temperature. The fatigue life is correlated with the maximum axial or circumferential strain range in
Z.B. Chen et al. / Materials Science and Engineering A 528 (2011) 8396–8401
8401
Fig. 11. Strain analysis of the coating in the circumferential direction for (a) IP TMF and (b) OP TMF.
Table 3 Circumferential stress in the coating. Applied mechanical strain (%) 0.50 0.45 0.40 0.35
Circumferential stress (MPa) IP
OP
−200.91 −178.71 −156.51 −132.92
−289.41 −272.29 −238.04 −220.91
the specimen and not just with the axial strain range. Further, it was published [10] that the IP TMF life was longer than the OP TMF life because cracking in the substrate rather than cracking in the bond coat or delamination at thermally grown oxide (TGO) controlled the OP TMF life. 4.2. Cracking behavior of the coating Actually, the state of strain in the coating is biaxial. The thermal mismatch strain is equal in all directions [6] and there is also a circumferential part of the applied mechanical strain which is −εmech (where is the Poisson ratio of the coating). The total circumferential strain on the coating, i.e. the sum of the thermal mismatch strain and the circumferential part of the applied mechanical strain, is thought to be a dominating factor for the coating delamination since it is directly related to the applied circumferential stress. For IP TMF cycling, the thermal mismatch strain is also positive but the circumferential part of the applied mechanical strain is negative. Thus the total circumferential strain range is smaller as shown in Fig. 11(a). However, for OP TMF cycling, the thermal mismatch strain is compressive and the circumferential part of the applied mechanical strain is also compressive. Consequently the total circumferential strain range is larger (Fig. 11(b)). The circumferential stress in the coating can be obtained when the circumferential strain is multiplied by the elastic modulus as shown in Table 3. It is seen that the stress in the coating under OP TMF is higher than that under IP TMF at the same applied mechanical strain. The higher stress in the circumferential direction is suggested to cause the delamination and spallation of the coating as observed. However, in the case of a PVD TBC the larger circumferential total mechanical strain range controlled the spalling behavior [6]. The plastic deformation from the interior of the specimen or from the severe fragmentation of the bond coat was also found to induce the detachment of the coating [10].
(1) The cyclic stress response is unsymmetrical. The mean stress is compressive in IP TMF tests but tensile in OP TMF tests. (2) Under the same mechanical strain amplitude, IP TMF life is longer than the OP TMF life because of the lower axial stress experienced under the IP condition. (3) The interface damage and cracking behavior are different in the two phase conditions. The specimens failed at the gauge length without delamination under the IP condition while the coating cracked and spalled in OP tests due to higher circumferential stresses acting on the coating. Acknowledgements The authors are very grateful to Prof. C. Laird for his English corrections of the manuscript. This work was financially supported by the Center for Interfacial Materials, Shenyang National Laboratory for Materials Science, Institute of Metal Research, Chinese Academy of Sciences. References [1] [2] [3] [4]
[5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19]
5. Conclusions
[20]
In the present work, the fracture behavior of an APS TBC system was studied under thermomechanical fatigue. The main conclusions are summarized below:
[21]
R.A. Miller, Surf. Coat. Technol. 30 (1987) 1–11. W.J. Brindley, R.A. Miller, Adv. Mater. Process. 136 (1989) 29–33. N.P. Padture, M. Gell, E.H. Jordan, Science 296 (2002) 280–284. U. Schulz, C. Leyens, K. Fritscher, M. Peters, B. Saruhan-Brings, O. Lavigne, J.M. Dorvaux, M. Poulain, R. Mévrel, M. Caliez, Aerosp. Sci. Technol. 7 (2003) 73–80. T.C. Totemeier, W.F. Gale, J.E. King, Metall. Mater. Trans. A 27 (1996) 363–369. P.K. Wright, Mater. Sci. Eng. A 245 (1998) 191–200. M. Bartsch, G. Marci, K. Mull, C. Sick, Adv. Eng. Mater. 1 (1999) 127–129. Y.H. Zhang, P.J. Withers, M.D. Fox, D.M. Knowles, Mater. Sci. Technol. Lond. 15 (1999) 1031–1036. E. Tzimas, H. Müllejans, S.D. Peteves, J. Bressers, W. Stamm, Acta Mater. 48 (2000) 4699–4707. B. Baufeld, E. Tzimas, P. Hähner, H. Müllejans, S.D. Peteves, P. Moretto, Scripta Mater. 45 (2001) 859–865. B. Baufeld, E. Tzimas, H. Müllejans, S. Peteves, J. Bressers, W. Stamm, Mater. Sci. Eng. A 315 (2001) 231–239. A. Peichl, T. Beck, O. Vöhringer, Surf. Coat. Technol. 162 (2003) 113–118. B. Baufeld, M. Bartsch, S. Dalkilic¸, M. Heinzelmann, Surf. Coat. Technol. 200 (2005) 1282–1286. J. Shi, A.M. Karlsson, B. Baufeld, M. Bartsch, Mater. Sci. Eng. A 434 (2006) 39–52. M. Bartsch, B. Baufeld, S. Dalkilic¸, L. Chernova, M. Heinzelmann, Int. J. Fatigue 30 (2008) 211–218. A.G. Evans, D.R. Mumm, J.W. Hutchinson, G.H. Meier, F.S. Pettit, Prog. Mater. Sci. 46 (2001) 505–553. W. Brandl, G. Marginean, D. Maghet, D. Utu, Surf. Coat. Technol. 188–189 (2004) 20–26. A. Scrivani, U. Bardi, L. Carrafiello, A. Lavacchi, F. Niccolai, G. Rizzi, J. Therm. Spray Technol. 12 (2003) 504–507. Z.W. Huang, Z.G. Wang, S.J. Zhu, F.H. Yuan, F.G. Wang, Mater. Sci. Eng. A 432 (2006) 308–316. L.Z. He, Q. Zheng, X.F. Sun, G.C. Hou, H.R. Guan, Z.Q. Hu, Mater. Sci. Eng. A 380 (2004) 340–348. C. Zhou, N. Wang, H. Xu, Mater. Sci. Eng. A 452–453 (2007) 569–574.
Contents lists available at SciVerse ScienceDirect
Materials Science and Engineering A journal homepage: www.elsevier.com/locate/msea
Thermomechanical fatigue behavior of an air plasma sprayed thermal barrier coating system Z.B. Chen a,c , Z.G. Wang a,∗ , S.J. Zhu b a b c
Shenyang National Laboratory for Materials Science, Institute of Metal Research, Chinese Academy of Sciences, 72 Wenhua Road, Shenyang 110016, PR China Department of Intelligent Mechanical Engineering, Fukuoka Institute of Technology, Higashi-ku, Fukuoka 811-0295, Japan Shenyang Aeroengine Research Institute, Aviation Industry Corporation of China, 1 Wanlian Road, Shenyang 110015, PR China
a r t i c l e
i n f o
Article history: Received 7 December 2010 Received in revised form 22 June 2011 Accepted 19 August 2011 Available online 25 August 2011 Keywords: Thermal barrier coating Thermomechanical fatigue Air plasma spraying
a b s t r a c t Failure behavior of an air plasma sprayed thermal barrier coating (TBC) system was investigated under inphase (IP) and out-of-phase (OP) thermomechanical fatigue (TMF) tests. All the TMF tests were performed in the temperature range of 450–850 ◦ C with a given period of 300 s under mechanical strain control. Both the bond coat NiCrAlY and the top coat 7%Y2 O3 –ZrO2 were fabricated by air plasma spraying (APS). Results revealed that the IP TMF lifetime was longer than that of the OP TMF under the same mechanical strain amplitude. Morphology observations of the failed specimens showed that the coating cracking and spallation processes were different in the two phase conditions. © 2011 Elsevier B.V. All rights reserved.
1. Introduction Thermal barrier coating (TBC) systems have been widely used to hot section parts of gas turbine engines to provide thermal protection to metallic components [1–3]. They offer potential for lowering the temperature of the metal substrate and thus produce an increase in power efficiency and a decrease in greenhouse gas emission. A typical TBC system consists of a superalloy substrate, a bond coat and a ceramic top coat, usually ZrO2 stabilized with 6–8 wt% Y2 O3 [4]. In service, hot section components experience severe cyclic temperature gradients and mechanical loads. As a consequence, thermomechanical fatigue (TMF), which provides a closer simulation of the actual strain-temperature cycle in an engine environment, is a major life limiting factor for gas turbine blades [5]. Therefore, TMF tests are useful for evaluating the service lifetime of a TBC system, for identifying the damage mechanism, and for a life modeling approach. However, only a few publications are available in the open literature with regard to TMF tests of TBC systems [6–15]. The main reasons for the limited number of publications are: (1) the difficulty of obtaining sufficiently powerful heating and cooling technique associated with the poor heat absorption and conductivity of the ZrO2 top coat in the TBC system [9,11], and (2) establishing a service-like temperature gradient across the thickness of the TBC system.
∗ Corresponding author. Tel.: +86 24 8397 8870; fax: +86 24 2389 1320. E-mail address: [email protected] (Z.G. Wang). 0921-5093/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2011.08.031
Different heating equipments have been adopted in TMF tests for TBC systems, such as radiation furnaces [6,12], direct or indirect induction heating [8–11], and lamp furnaces [7,13–15]. The first published paper reported the results for a silicon carbide igniter furnace for heating a TBC system, which produced a temperature gradient around 50 ◦ C during transient heating and less than 20 ◦ C during a high temperature dwell [6]. Then induction heating was used to perform a TMF test on a TBC system and obtained an inverse temperature gradient across the coating that was contrary to the in-service situation [8]. Later, an indirect induction heating method was developed and produced a temperature difference of 50 ◦ C between the top coat and substrate during the high temperature hold, but the heat transfer susceptor was hard to select and the test period was long [9–11]. A heating system closer to the ideal was the lamp furnace, which could heat specimens up to 1000 ◦ C in a few seconds [13–15] and obtain a temperature gradient of about 170 ◦ C [14,15]. Among these studies, it was found that the relationship of the lifetime for TBC systems between in-phase (IP) TMF and out-of-phase (OP) TMF is variable. Wright [6] found that the OP TMF lifetime of the TBC system was longer than the IP TMF one, whereas Baufeld et al.’s experiments [10] revealed that the IP TMF lifetime was longer. The cracking and failure mechanisms of TBC systems reported by different authors were also variable. Although TBC systems typically fail by coating delamination or spallation under TMF conditions [6,8–11,13,15], the crack initiation sites are different. Cracks may develop from the uncoated inner surface [10,11], the bond coat [8,10], or the interface between the bond coat and the top coat [9]. As a result, it is difficult to understand the fracture behavior of TBC systems under service conditions and to predict the lifetime of the components.
Z.B. Chen et al. / Materials Science and Engineering A 528 (2011) 8396–8401
8397
Fig. 1. Dimensions of TBC specimens for thermomechanical fatigue tests.
The bond coat, usually MCrAlY overlay coating (where M is Ni, Co, or a combination of these elements), is the most crucial component of the TBC system [16] that can be produced by thermal spraying, sputtering or evaporation; however, for practical applications, thermally sprayed coatings are normally preferred [17]. Among thermal spraying techniques, APS and high velocity oxygen fuel (HVOF) can be conducted at atmospheric pressure, therefore their equipment investment and production costs are low compared with vacuum plasma spraying (VPS). Meanwhile, APS shows a considerably better efficiency than other methods [18]. Thus it is attractive to spray the bond coat by APS. The aim of this paper is to present experimental results concerning TMF lifetime and failure behavior of a TBC system consisting of a Ni-based superalloy substrate, a NiCrA1Y bond coat and a Y2 O3 –ZrO2 top coat, both of which were fabricated by APS. The relationship of the lifetime and differences of the cracking behaviors between the IP and OP conditions are discussed. 2. Experimental The cast Ni-based superalloy M963 was used as a substrate. Its chemical composition (in wt.%) was C 0.15, Cr 8.89, Al 6.00, Ti 2.55, Mo 1.64, W 10.1, Co 10.0, Nb 1.10, Zr 0.03, B 0.03, Ce 0.02, Y 0.01 and Ni balance. Before machining, the specimens were solution-treated at 1210 ◦ C for 4 h followed by air cooling. Cylindrical tube specimens were machined with a total length of 135 mm, an internal diameter of 7 mm and in the gauge length an external diameter of 10 mm as shown in Fig. 1. Before spraying these specimens, all the substrates were grit blasted by alumina powder with 80 mesh grain size distribution. The coating substrate was deposited by APS with a Ni–25Cr–5Al–0.5Y alloy as the bond coat with a normal thickness of 130 m and 7%Y2 O3 –ZrO2 as the top coat with 250 m thickness. The APS setting was METCO 7 M and spray parameters for the bond coat and the top coat are listed in Table 1. The TMF tests were performed on an MTS810 closed-loop servohydraulic testing machine with computer control. A radiation furnace powered by four cylindrical quartz lamps, each with a maximum power of 2.5 kW, was used for heating. The radiation of the quartz lamps was focused on the specimen by reflection from elliptical mirrors. The outer surface temperature of the specimen, which was controlled by the power output of the lamps, was measured with a thermocouple enlacing the specimen. Cooling was obtained by internally compressed air combined with reducing the power output of the lamps. Axial strain amplitudes were controlled using a self-supporting extensometer that had
a gauge length of 23 mm, supported with ceramic rods. A triangle waveform was used for both thermal cycling and mechanical cycling. Two kinds of TMF tests were used: IP where the maximum mechanical strain coincided with the maximum temperature and OP where the maximum mechanical strain was attained at the minimum temperature. TMF tests were carried out in the temperature range of 450–850 ◦ C with a cyclic period of 300 s under mechanical strain control. The mechanical strain amplitudes were varied from 0.35% to 0.5% with a strain ratio of −1. The apparatus yielded a temperature difference of about 90 ◦ C between the top coat and the substrate during dwells at the top coat temperature of 900 ◦ C. After the TMF tests, the cyclically induced morphologies of all specimens, were recorded by a high accuracy digital camera, and fracture surfaces were investigated using a scanning electron microscope (SEM). Longitudinal sections were cut from the tested specimens and were embedded in an epoxy resin. Then these sections were metallographically prepared and observed by the SEM. 3. Results 3.1. Cyclic deformation and fatigue lifetime Fig. 2 shows the typical stress–mechanical strain hysteresis loops for IP TMF and OP TMF at a mechanical strain amplitude of 0.45%. It can be seen that the maximum and minimum stress of each hysteresis loop are unsymmetrical. In IP TMF, the value of the maximum stress is lower than that of the minimum one, while it is opposite under OP TMF, and the value of the maximum stress is higher. The cyclic stress response behavior at a mechanical strain amplitude of 0.45% is shown in Fig. 3, which displays the variation of the maximum stress ( max ) and the minimum stress ( min ) as well as the mean stress ( m ) with the number of cycles (N). In the tensile stress cycles, the maximum stress ( max ) gradually decreases in IP TMF but increases under OP TMF with an increase in N. In the compressive stress cycles, the tendency of the minimum stress ( min )
Table 1 Spray parameters for bond coat and top coat. Bond coat Spraying distance (mm) Feed rate (g/min) Carrier gas (Ar) pressure (MPa) Voltage (V) Current (A)
145–150 30–45 0.40–0.55 40–60 500–530
Top coat 70–120 35–45 0.40–0.55 45–65 500–550
Fig. 2. Middle-life hysteresis loops for IP and OP TMF at εmech /2 = 0.45%.
8398
Z.B. Chen et al. / Materials Science and Engineering A 528 (2011) 8396–8401
The hysteresis loops and cyclic stress response behaviors of the TBC system are similar to the corresponding material of the substrate M963, while the relationship of the fatigue life for the TBC system between the IP and OP conditions is different from that of the substrate [19]. For the bare substrate, IP TMF and OP TMF fatigue lives with regard to the mechanical strain amplitude exhibit a crossover. The IP TMF life is shorter than the OP TMF life at high mechanical strain amplitudes whereas it is longer than the OP TMF life at low mechanical strain amplitudes, which is mainly due to creep and oxidation variations of the specimens. However, creep and oxidation do not influence the lifetime of the superalloy with TBC because the coating prevents the substrate from damage at high temperature in the TBC system; moreover, the highest temperature in this test program was just 850 ◦ C. 3.2. Morphology observation Fig. 3. Cyclic stress response curves for IP and OP TMF at εmech /2 = 0.45%.
Fig. 4. Mechanical strain amplitude as a function of cycles to failure.
is just opposite. As a result, the mean stress ( m ) is compressive in IP TMF but tensile in OP TMF. The fatigue life for IP and OP TMF as a function of mechanical strain amplitude is shown in Fig. 4. It can be seen that the fatigue life decreases with an increase in the mechanical strain amplitude. Under the same mechanical strain amplitude, OP TMF exhibits a shorter fatigue life than IP TMF.
Typical photos of failed specimens recorded by a digital camera are shown in Fig. 5. It can be seen that the failure behaviors of the specimens are different between the two phase conditions. The specimens failed in the gauge lengths in the IP tests for all four mechanical strain amplitudes (Fig. 5(a)); however, under the OP condition, at all the four mechanical strain amplitudes, there was cracking in the coating though the specimens also failed within the gauge lengths (Fig. 5(b)). From the inset of Fig. 5(b) it is shown that the coating spallation was located at the interface of the bond coat/substrate. A representative fractograph of failed specimens under the IP TMF condition is shown in Fig. 6(a). No crack is found in the coating and the interfaces are almost intact. By contrast, the fracture surface of the specimen failed in the OP TMF test shows a different failure behavior in that the bond coat is widely detached from the substrate (Fig. 6(b)). The fracture surfaces are consistent with the appearance of the TMF failed specimens as shown in Fig. 5. Typical longitudinal sections close to a fracture surface are shown in Fig. 7. No obvious delamination can be seen in the IP TMF tests as shown in Fig. 7(a). A crack is detected only in the top coat as indicated by the arrow. However, there is a crack passing through both the top coat and the bond coat under the OP TMF condition (Fig. 7(b)). The crack does not propagate into the substrate but is deflected along the interface between the bond coat and the substrate, causing interface delamination. Cracks can also be found at the interface of the top coat/bond coat. The phenomena observed from the longitudinal sections are in accordance with the macroscopic surface and fracture surface observations.
Fig. 5. Photos of failed specimens at (a) IP TMF and (b) OP TMF, inset of (b) showing spallation of the coating.
Z.B. Chen et al. / Materials Science and Engineering A 528 (2011) 8396–8401
8399
Fig. 6. Fractographs of failed specimens at (a) IP TMF and (b) OP TMF (TC, top coat; BC, bond coat; Sub, substrate).
Fig. 7. SEM images of longitudinal sections of fracture specimens at (a) IP TMF and (b) OP TMF.
neglected. Besides the applied mechanical strain, there is a thermal mismatch strain arising from the thermal expansion mismatch between the coating and substrate during the TMF testing. The thermal expansion of the coating is taken to be that of the bond coat because the damage occurred at the bond coat/substrate interface. The thermal strain εth of the coating or the substrate is calculated as follows: εth = ˛ T
(1)
where ˛ is the mean coefficient of thermal expansion in the temperature range of T. The specimen usually fails at the maximum tensile stress which corresponds to the maximum and minimum temperature for IP TMF or OP TMF, respectively. Therefore, the maximum thermal mismatch strain range εth between the coating and the substrate in the TMF tensile cycle can be calculated as: Fig. 8. Sketch of strains in a coated specimen.
εth = ˛M963 (Tmax − Tmiddle ) − ˛BC (Tmax − Tmiddle )
(2.1)
εth = ˛M963 (Tmin − Tmiddle ) − ˛BC (Tmin − Tmiddle )
(2.2)
4. Discussion According to the experimental results, the TBC system can be considered as a two part composite: the substrate, and the combination of the top coat with the bond coat. A simple model is proposed as shown in Fig. 8 where the coating consists of the bond coat and the top coat. For simplicity, the coating and substrate are assumed to experience the same applied mechanical strain; effects of the interface bonding strength on the cracking behaviors are
Eq. (2.1) applies to the IP TMF test and the maximum thermal mismatch strain range arrives at the maximum temperature 850 ◦ C. Eq. (2.2) applies to the OP TMF test and the maximum thermal mismatch strain range obtains at the minimum temperature 450 ◦ C. Tmiddle is the mid temperature of the test, which is 650 ◦ C. The parameters such as elastic modulus, coefficient of thermal expansion and Poisson ratio are shown in Fig. 9(a)–(c), respectively [20,21].
8400
Z.B. Chen et al. / Materials Science and Engineering A 528 (2011) 8396–8401
Fig. 9. (a) Elastic modulus and (b) coefficient of thermal expansion as a function of temperature for substrate and bond coat; (c) Poisson ratio as a function of temperature for bond coat [20,21].
Fig. 10. Strain analysis of the coating in the axial direction for (a) IP TMF and (b) OP TMF.
4.1. Relationship of fatigue lifetime to IP TMF and OP TMF The total axial strain on the coating, i.e. the sum of the applied mechanical strain and the thermal mismatch strain, is a crucial factor for the coating life since it is directly related to the applied axial stress. During IP TMF cycling, the temperature increases from the mid-range to the maximum, and the thermal mismatch strain of the coating is positive from Eq. (2.1) because the coefficient of thermal expansion of the coating is lower than that of the substrate. The thermal strain adds to the applied mechanical strain, producing a larger total axial strain range as shown in Fig. 10(a). Whereas during OP TMF cycling, the temperature decreases from the mid-range to the minimum, and the thermal mismatch strain of the coating is negative in Eq. (2.2), which is of opposite sign to that of the applied mechanical strain. Therefore the total axial strain obtains a lower strain range (Fig. 10(b)). However, the elastic moduli of the coating differ between the maximum and minimum temperature. When the total axial strain is multiplied by the elastic modulus, the experienced axial stress can be calculated as shown in Table 2. It
Table 2 Axial stress in the coating. Applied mechanical strain (%) 0.50 0.45 0.40 0.35
Axial stress (MPa) IP
OP
716.23 646.85 577.48 508.10
823.71 738.09 652.46 566.84
is found that the stress in the coating under OP TMF is higher than that under IP TMF at the same applied mechanical strain. Therefore, a shorter fatigue lifetime under OP TMF is expected. The differences of fatigue lifetime between IP and OP TMF found here is in conflict with results reported in the open literature. For example, it was reported [6] that OP TMF life was longer than IP TMF life, which was believed to be governed by the extent of compression at minimum cycle temperature. The fatigue life is correlated with the maximum axial or circumferential strain range in
Z.B. Chen et al. / Materials Science and Engineering A 528 (2011) 8396–8401
8401
Fig. 11. Strain analysis of the coating in the circumferential direction for (a) IP TMF and (b) OP TMF.
Table 3 Circumferential stress in the coating. Applied mechanical strain (%) 0.50 0.45 0.40 0.35
Circumferential stress (MPa) IP
OP
−200.91 −178.71 −156.51 −132.92
−289.41 −272.29 −238.04 −220.91
the specimen and not just with the axial strain range. Further, it was published [10] that the IP TMF life was longer than the OP TMF life because cracking in the substrate rather than cracking in the bond coat or delamination at thermally grown oxide (TGO) controlled the OP TMF life. 4.2. Cracking behavior of the coating Actually, the state of strain in the coating is biaxial. The thermal mismatch strain is equal in all directions [6] and there is also a circumferential part of the applied mechanical strain which is −εmech (where is the Poisson ratio of the coating). The total circumferential strain on the coating, i.e. the sum of the thermal mismatch strain and the circumferential part of the applied mechanical strain, is thought to be a dominating factor for the coating delamination since it is directly related to the applied circumferential stress. For IP TMF cycling, the thermal mismatch strain is also positive but the circumferential part of the applied mechanical strain is negative. Thus the total circumferential strain range is smaller as shown in Fig. 11(a). However, for OP TMF cycling, the thermal mismatch strain is compressive and the circumferential part of the applied mechanical strain is also compressive. Consequently the total circumferential strain range is larger (Fig. 11(b)). The circumferential stress in the coating can be obtained when the circumferential strain is multiplied by the elastic modulus as shown in Table 3. It is seen that the stress in the coating under OP TMF is higher than that under IP TMF at the same applied mechanical strain. The higher stress in the circumferential direction is suggested to cause the delamination and spallation of the coating as observed. However, in the case of a PVD TBC the larger circumferential total mechanical strain range controlled the spalling behavior [6]. The plastic deformation from the interior of the specimen or from the severe fragmentation of the bond coat was also found to induce the detachment of the coating [10].
(1) The cyclic stress response is unsymmetrical. The mean stress is compressive in IP TMF tests but tensile in OP TMF tests. (2) Under the same mechanical strain amplitude, IP TMF life is longer than the OP TMF life because of the lower axial stress experienced under the IP condition. (3) The interface damage and cracking behavior are different in the two phase conditions. The specimens failed at the gauge length without delamination under the IP condition while the coating cracked and spalled in OP tests due to higher circumferential stresses acting on the coating. Acknowledgements The authors are very grateful to Prof. C. Laird for his English corrections of the manuscript. This work was financially supported by the Center for Interfacial Materials, Shenyang National Laboratory for Materials Science, Institute of Metal Research, Chinese Academy of Sciences. References [1] [2] [3] [4]
[5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19]
5. Conclusions
[20]
In the present work, the fracture behavior of an APS TBC system was studied under thermomechanical fatigue. The main conclusions are summarized below:
[21]
R.A. Miller, Surf. Coat. Technol. 30 (1987) 1–11. W.J. Brindley, R.A. Miller, Adv. Mater. Process. 136 (1989) 29–33. N.P. Padture, M. Gell, E.H. Jordan, Science 296 (2002) 280–284. U. Schulz, C. Leyens, K. Fritscher, M. Peters, B. Saruhan-Brings, O. Lavigne, J.M. Dorvaux, M. Poulain, R. Mévrel, M. Caliez, Aerosp. Sci. Technol. 7 (2003) 73–80. T.C. Totemeier, W.F. Gale, J.E. King, Metall. Mater. Trans. A 27 (1996) 363–369. P.K. Wright, Mater. Sci. Eng. A 245 (1998) 191–200. M. Bartsch, G. Marci, K. Mull, C. Sick, Adv. Eng. Mater. 1 (1999) 127–129. Y.H. Zhang, P.J. Withers, M.D. Fox, D.M. Knowles, Mater. Sci. Technol. Lond. 15 (1999) 1031–1036. E. Tzimas, H. Müllejans, S.D. Peteves, J. Bressers, W. Stamm, Acta Mater. 48 (2000) 4699–4707. B. Baufeld, E. Tzimas, P. Hähner, H. Müllejans, S.D. Peteves, P. Moretto, Scripta Mater. 45 (2001) 859–865. B. Baufeld, E. Tzimas, H. Müllejans, S. Peteves, J. Bressers, W. Stamm, Mater. Sci. Eng. A 315 (2001) 231–239. A. Peichl, T. Beck, O. Vöhringer, Surf. Coat. Technol. 162 (2003) 113–118. B. Baufeld, M. Bartsch, S. Dalkilic¸, M. Heinzelmann, Surf. Coat. Technol. 200 (2005) 1282–1286. J. Shi, A.M. Karlsson, B. Baufeld, M. Bartsch, Mater. Sci. Eng. A 434 (2006) 39–52. M. Bartsch, B. Baufeld, S. Dalkilic¸, L. Chernova, M. Heinzelmann, Int. J. Fatigue 30 (2008) 211–218. A.G. Evans, D.R. Mumm, J.W. Hutchinson, G.H. Meier, F.S. Pettit, Prog. Mater. Sci. 46 (2001) 505–553. W. Brandl, G. Marginean, D. Maghet, D. Utu, Surf. Coat. Technol. 188–189 (2004) 20–26. A. Scrivani, U. Bardi, L. Carrafiello, A. Lavacchi, F. Niccolai, G. Rizzi, J. Therm. Spray Technol. 12 (2003) 504–507. Z.W. Huang, Z.G. Wang, S.J. Zhu, F.H. Yuan, F.G. Wang, Mater. Sci. Eng. A 432 (2006) 308–316. L.Z. He, Q. Zheng, X.F. Sun, G.C. Hou, H.R. Guan, Z.Q. Hu, Mater. Sci. Eng. A 380 (2004) 340–348. C. Zhou, N. Wang, H. Xu, Mater. Sci. Eng. A 452–453 (2007) 569–574.