- Email: [email protected]
Laser cycling and thermal cycling exposure of thermal barrier coatings on copper substrates
Surface & Coatings Technology 206 (2011) 1605–1608
Contents lists available at SciVerse ScienceDirect
Surface & Coatings Technology j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s u r f c o a t
Laser cycling and thermal cycling exposure of thermal barrier coatings on copper substrates Jana Schloesser ⁎, Martin Bäker, Joachim Rösler Technische Universität Braunschweig, Institut für Werkstoffe, Langer Kamp 8, 38106 Braunschweig, Germany
a r t i c l e
i n f o
Available online 16 August 2011 Keywords: Thermal barrier coating Copper Rocket engine Laser shock testing
a b s t r a c t Thermal barrier coatings (TBCs) are successfully applied in turbines and could also protect combustion chambers in rocket engines. Apart from different loading conditions, the main difference between these applications is the substrate material, which is nickel-based for turbines and copper-based for rocket engines. To optimize the coating system, more knowledge of possible failure modes is necessary. In this work a standard coating system was applied by atmospheric plasma spraying to copper specimens. These specimens were exposed to thermal cycling with different cooling rates and to laser shock testing. A laser-cycling set-up was developed to qualify different coating systems. This set-up consists of a high-power diode laser (3 kW) which provides high heating rates to up to 1500 °C. Laser shock testing has proven to be a suitable alternative to burner rig testing. The results were different to the common failure modes for TBCs on nickel substrates as the coatings system does not fail at the interface between top coat and bond coat, but at the interface between substrate and bond coat. Two failure modes were observed: copper oxide was undermining the coatings at the substrate/bond coat-interface in the case of thermal cycling experiments, and complete delamination occurred at the same interface in the case of laser shock testing. Consequently, this interface is critical in the investigated material system. © 2011 Elsevier B.V. All rights reserved.
1. Introduction Thermal barrier coatings (TBCs) are effectively used in turbine applications. A TBC usually consists of a metallic bond coat and a ceramic top coat and provides thermal as well as oxidation protection. The concept of a TBC could also be transferred to rocket engines. Particularly the combustion chamber experiences extremely high thermal loads during rocket launch. To realize relaunchable space transportation systems, it is important to enhance the lifetime of the combustion chamber. After several launches the thin copper cooling channels of the chamber fail because of the so-called doghouse effect [1] which is a combination of creep, thermo-mechanical fatigue, and environmental attack. Loading conditions in a rocket engine are different compared to gas turbine application. The loading cycles are very short and only about 55 cycles are required [2]. Therefore, the total accumulated service time at high temperature of the combustion liner is expected to be in the range of 2 h [3]. Maximum service temperature depends on the cooling efficiency and on the thickness and thermal conductivity of the applied coating. For uncoated copper temperatures are expected to be about 600 °C [4]. Because of the high gas
⁎ Corresponding author. E-mail address: [email protected] (J. Schloesser). 0257-8972/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.surfcoat.2011.07.084
temperatures and the liquid hydrogen (LH2) cooling inside the cooling channels, high thermal gradients are imposed and heat fluxes are expected to be in the range of 160 MW/m 2 [5]. To protect the material from the oxidizing environment and from high thermal loads, a protective coating could be applied. Although the thermo mechanical loading conditions differ, the main requirements of a thermal barrier coating (thermal and oxidation protection) are the same as in other applications. While first investigations were already published in the 80s (e.g. by Quentmeyer et al. [6]), several coating systems have also been considered in recent research. A thermal barrier coating system consisting of a NiCrAlY top coat and a Cu\Cr bond coat on Cu–8%Cr4%Nb-substrate has been investigated by Jain et al. [7] and Raj et al. [2,4]. These coatings were coldsprayed and hot-isostatically pressed and showed good adhesion in the initial thermal cycling tests. A TBC-system consisting of a MCrAlY-bond coat and zirconia was applied by electron beam vapor deposition by Schulz et al. [8]. This concept for TBC optimization on copper substrates starts from a well-known standard TBC system for nickel alloys. Although optimized for a nickel based substrate, a well-characterized coating system was chosen to investigate mainly the influence of the different loading conditions and the substrate. To enhance the lifetime of a suitable coating on copper substrates it is essential to know more about the failure mechanisms occurring in this new application. Unfortunately not much is published about the exact failure mechanisms.
1606
J. Schloesser et al. / Surface & Coatings Technology 206 (2011) 1605–1608
Due to the difference in materials' properties of copper-based (rocket engine) and nickel-based (turbines) alloys and to the difference in loading conditions, it is not clear whether the known failure mechanisms of thermal barrier coatings in gas turbines [9] can be transferred to rocket engines. Therefore, state-of-the-art thermal barrier coatings were applied to copper samples and exposed to thermal cycling and laser shock testing to investigate possible failure mechanisms. 2. Experimental procedures 2.1. Investigated materials Due to its good high temperature properties, the commercial alloy CuCrZr (specification number 2.1293), which is an age-hardenable copper alloy containing 1% chromium and 0.3% zirconium [10], was chosen as the substrate material. A standard coating system know from gas turbine applications was applied by atmospheric plasma spraying (APS) to the specimens. It consisted of a NiCrAlY bond coat (Ni–22%Cr–10%A–1%Y) and a partially stabilized zirconia (8% yttrium oxide) top coat. The coating system was chosen because it is already well known and characterized, although optimized for nickel based substrates. The coating parameters were optimized for the new substrate material [11], resulting in a well adhering coating.
been practiced in the past [12–15]. The advantage of this method is the well-defined heat flux and the contact-free heating compared to usual burner-rig-testing (an overview is given in [16]), that is commonly used to assess the lifetime of TBCs. The laser shock testing reported in literature was mainly performed with a moving laser spot (e.g. [15]). To realize a more homogeneous temperature distribution, a new set-up with a broad focal point was used in the present study (see Fig. 1). This is especially important as the copper substrate has a good heat conductivity. The diode laser used has a maximum power of 3 kW and the whole specimen surface can be heated homogenously. The surface temperature is controlled by a high speed pyrometer. Surface temperatures of up to 1500 °C can be reached within 0.2 s with the laser shock set-up while the backside of the samples is cooled by air. A surface temperature versus time plot is shown in Fig. 2. The plot also shows that the surface temperature is kept stable during the heating time of 2 s. Thin disk-shaped specimens with a thickness of 2 mm and a diameter of 20 mm were coated for laser shock testing with bond coat and thermal barrier coating. Zirconia has a poor absorption for the wavelengths used [17]. To ensure good absorption, the TBC was treated with a Fe3O4 powder suspension before testing. 3. Results and discussion
2.2. Thermal cycling experiments
3.1. Failure produced by thermal cycling
Thermal cycling experiments were performed on flat specimens with 35 mm width, 30 mm height and 5 mm thickness. One face of the specimens was completely coated (30 mm × 35 mm) while the other faces remained uncoated. The cycling experiments were performed to assess the adherence of the produced coatings and to examine the failure mechanisms. The specimens were heated up to 800 °C inside an oven, and, after a holding time of 30 min, they were cooled down in different cooling agents outside the oven. This cycle was repeated at least 50 times on each specimen in case of no failure of the coating system. The described cycling method is “isothermal” in the sense that all parts of the coating and the substrate are at the same temperature at the same time and are cooled and heated more or less simultaneously. No thermal gradient was applied to the coating. The temperature was also limited by the copper substrate. Nevertheless, due to the difference in the coefficients of thermal expansion, large thermal strains are imposed during heating and cooling of the specimen, which could lead to a spallation of the coating. The specimens were optically inspected after each cycle and after each 5–10 cycles one specimen was taken from the experiment for metallographic preparation. The prepared samples were investigated by optical microscopy as well as by scanning electron microscopy (SEM). To examine the effect of cooling rates, three different cooling agents were used: air, icy water and liquid nitrogen (LN2). The cooling rates were measured with a calibration specimen equipped with thermocouples. Air cooling shows the slowest cooling rate with about 0.5 K/s from 800 °C to room temperature. The fastest cooling rate, 13 K/s from 800 °C to 10 °C, is reached with icy water. Liquid nitrogen shows an in-between value with 7 K/s from 800 °C to −140 °C. Cooling with liquid nitrogen was applied because it reaches a low temperature (liquid hydrogen is used in real application), although due to gas formation at the surface of the specimen the cooling rate is not as fast as in icy water.
Specimens that were cooled down slowly (air cooling) did not show any failure of the thermal barrier coating after 50 cycles. A micrograph of a sample after exposure can be seen in Fig. 3. At the uncoated edges the copper oxidizes severely but the coating itself provides a certain oxidation protection. Due to the oxidation of the copper at the uncoated edges, an overhang of the coating was formed (Fig. 3). No spallation or partial delamination of the thermal barrier coating was observed with this cooling rate. At higher cooling rates (icy water and LN2) some of the specimens behaved similar to those cooled down slowly and did not show any sign of failure. Other samples showed partial delamination at the interface between substrate and bond coat (compare Fig. 4). A layered structure of copper oxide formed at these delaminated regions. Delamination started only at the uncoated edges where the copper oxide undermines the coating. A layer of copper oxide was formed in 1 cycle and then spalled during cooling, the next layer of copper oxide was then formed at high temperatures and the whole process was repeated. Raj et al. [2] observed a similar behavior in their oxidation tests at the uncoated sample edges, but in contrast to our results, the coatings were not undermined. GRCop-84 was in used in their study which has a better general oxidation resistance. It is believed that the substrate alloy plays a more important role concerning the susceptibility to this failure mode than the coating system. Statistical variations of the coatings are believed to be the reason why this phenomenon is seen only in part of the samples. This failure
2.3. Laser shock testing To account for the high thermal gradients in real service, a different testing method than thermal cycling is needed. Heating the samples on only one side by a laser is an approach that has already
Fig. 1. Set-up for laser shock testing.
J. Schloesser et al. / Surface & Coatings Technology 206 (2011) 1605–1608
1750
Surface temperature Set temperature
1500
Temperature [° C]
1607
1250 1000 750 500 250 0 0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
Time [s] Fig. 2. Surface temperature versus time for laser shock experiments showing high heating rates.
mechanism is critical under operational conditions because a totally dense coating cannot be assured due to erosion. This mechanism is mainly driven by oxidation of the substrate, and none of the coatings failed completely.
3.2. Failure produced by laser shock testing Laser cycling experiments have been performed at different surface temperatures from 750 °C and 1500 °C. A metallographic post mortem analysis was applied to each specimen. The specimens failed completely after the first cycle (2 s holding time at high temperatures) at temperatures above 1250 °C. With lower temperature, the samples survive more cycles. A failure plot for 2 s holding time is shown in Fig. 5. These results show that the thermal gradient plays an important role for the lifetime of the coatings as the thermal gradient is increasing with increasing surface temperature. Comparing the results to thermal cycling, the samples survived only 8 cycles at 750 °C compared with more than 50 cycles in thermal cycling experiments. This indicates clearly that it is not enough to test the coating systems with thermal cycling experiments without thermal gradients. Fig. 6 shows a typical failure after laser cycling. The thermal shock induced by the laser leads to a delamination of the whole coating at the substrate interface. In none of the experiments partial failure was observed. No sign of oxidation at the interface between substrate and bond coat was visible with scanning electron microscopy (SEM).
Fig. 4. Micrograph after 50 cycles (800 °C, LN2 cooling) showing partial delamination.
No formation of thermally grown oxide on the bond coat/substrate interface was observed. This was the case because the heating times were very short compared to normal thermal cycling test. But as the service times in rocket engines are also quite short it is not likely that a thermally grown oxide will form, which plays a crucial role in the failure mechanisms of thermal barrier coatings on nickel based substrates. Due to the large difference in thermal expansion coefficients of the bond coat and the substrate, large thermal stresses were probably induced in this region. These results show on the one hand that, again, the substrate interface is the weak interface and on the other hand that the laser heating method is a suitable method to qualify different coating systems because it produces damage and even delamination by thermal stresses. Therefore, it is an important tool for future coating development.
4. Conclusions Possible failure modes of TBCs on copper substrates were investigated in this paper. The observed failure modes differ significantly compared to specimens with nickel substrate (e.g. for gas turbine applications). Thermal cycling led to partial delamination due to copper oxide formation only at specimen edges for high cooling rates. The copper oxide was undermining the coating at the interface between bond coat and substrate. This failure mode is critical in case of a damaged layer e.g. due to erosion in a rocket engine. No complete spallation of the coatings occurred during thermal cycling. No formation of a thermally
2s holding time
Temperature [° C]
1500
1250
1000
750 no failure failure
500 0
2
4
6
8
10
Number of cycles Fig. 3. Micrograph after 50 cycles (800 °C, air cooling) showing no delamination.
Fig. 5. Cycles to failure depending on surface temperature (laser heating time 2 s).
1608
J. Schloesser et al. / Surface & Coatings Technology 206 (2011) 1605–1608
References
Fig. 6. Typical failure after laser cycling at the interface between substrate and bond coat.
grown oxide, which is critical in failure of thermal barrier coatings on gas turbines, was observed. To account for the thermal gradient inside the coating and fast heating rates, a laser shock experiment was set up to qualify the coatings. These experiments show delamination at the interface between substrate and bond coat. This failure mechanism will be investigated in subsequent studies. All experiments show that the interface between bond coat and the substrate is the critical interface for the investigated material system. For further thermal barrier coating development, the optimization of this interface is the main focus.
[1] J. Riccius, O. Haidn, E. Zametaev, 40th AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, 2004. [2] S.V. Raj, C. Barrett, J. Karthikeyan, R. Garlick, Surf. Coat. Technol. 201 (2007) 7222. [3] L.U. Ogbuji, Surf. Coat. Technol. 197 (2005) 327. [4] S.V. Raj, L.J. Ghosn, C. Robinson, D. Humphrey, Mater. Sci. Eng. A 457 (2007) 300. [5] S. Beyer, D. Klemm, M. Bobeth, W. Pompe, O. Trommer, P. Gawlitza, Adv. Eng. Mater. 7 (2005) 54. [6] R.J. Quentmeyer, H.J. Kasper, J.M. Kazaroff, AIANSAE 14th, Joint Propulsion Conference, 1978. [7] P. Jain, S.V. Raj, K.J. Hemker, Acta Mater. 55 (2007) 5103. [8] U. Schulz, K. Fritscher, M. Peters, D. Greuel, O. Haidn, Sci. Technol. Adv. Mater. 6 (2005) 103. [9] J. Rösler, M. Bäker, M. Volgmann, Acta Mater. 49 (2001) 3659. [10] Deutsches Kupferinstitut, Kupferdatenblatt CuCr1Zr, Deutsches Kupferinstitut, 2005. [11] J. Schloesser, T. Fedorova, M. Bäker, J. Rösler, J. Solid Mech. Mater. Eng. 4 (2010) 189. [12] W. Pompe, H.-A. Bahr, I. Pflugbeil, G. Kirchhoff, P. Langmeier, H.-J. Weiss, Mater. Sci. Eng. A 233 (1997) 167. [13] S. Rangaraj, K. Kokini, Acta Mater. 52 (2004) 455. [14] A.S. Semenov, H.-A. Bahr, H. Balke, H.-J. Weiss, Thin Solid Films 471 (2005) 200. [15] D. Nies, R. Pulz, S. Glaubitz, M. Finn, B. Rehmer, B. Skrotzki, Adv. Eng. Mater. 12 (2010) 1224. [16] R. Vaßen, F. Cernuschi, G. Rizzi, A. Scrivani, N. Markocsan, L. Östergren, A. Kloosterman, R. Mevrel, J. Feist, J. Nicholls, Adv. Eng. Mater. 10 (2008) 907. [17] J. Eldridge, C. Spuckler, J. Am. Ceram. Soc. 91 (2008) 1603.
Contents lists available at SciVerse ScienceDirect
Surface & Coatings Technology j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s u r f c o a t
Laser cycling and thermal cycling exposure of thermal barrier coatings on copper substrates Jana Schloesser ⁎, Martin Bäker, Joachim Rösler Technische Universität Braunschweig, Institut für Werkstoffe, Langer Kamp 8, 38106 Braunschweig, Germany
a r t i c l e
i n f o
Available online 16 August 2011 Keywords: Thermal barrier coating Copper Rocket engine Laser shock testing
a b s t r a c t Thermal barrier coatings (TBCs) are successfully applied in turbines and could also protect combustion chambers in rocket engines. Apart from different loading conditions, the main difference between these applications is the substrate material, which is nickel-based for turbines and copper-based for rocket engines. To optimize the coating system, more knowledge of possible failure modes is necessary. In this work a standard coating system was applied by atmospheric plasma spraying to copper specimens. These specimens were exposed to thermal cycling with different cooling rates and to laser shock testing. A laser-cycling set-up was developed to qualify different coating systems. This set-up consists of a high-power diode laser (3 kW) which provides high heating rates to up to 1500 °C. Laser shock testing has proven to be a suitable alternative to burner rig testing. The results were different to the common failure modes for TBCs on nickel substrates as the coatings system does not fail at the interface between top coat and bond coat, but at the interface between substrate and bond coat. Two failure modes were observed: copper oxide was undermining the coatings at the substrate/bond coat-interface in the case of thermal cycling experiments, and complete delamination occurred at the same interface in the case of laser shock testing. Consequently, this interface is critical in the investigated material system. © 2011 Elsevier B.V. All rights reserved.
1. Introduction Thermal barrier coatings (TBCs) are effectively used in turbine applications. A TBC usually consists of a metallic bond coat and a ceramic top coat and provides thermal as well as oxidation protection. The concept of a TBC could also be transferred to rocket engines. Particularly the combustion chamber experiences extremely high thermal loads during rocket launch. To realize relaunchable space transportation systems, it is important to enhance the lifetime of the combustion chamber. After several launches the thin copper cooling channels of the chamber fail because of the so-called doghouse effect [1] which is a combination of creep, thermo-mechanical fatigue, and environmental attack. Loading conditions in a rocket engine are different compared to gas turbine application. The loading cycles are very short and only about 55 cycles are required [2]. Therefore, the total accumulated service time at high temperature of the combustion liner is expected to be in the range of 2 h [3]. Maximum service temperature depends on the cooling efficiency and on the thickness and thermal conductivity of the applied coating. For uncoated copper temperatures are expected to be about 600 °C [4]. Because of the high gas
⁎ Corresponding author. E-mail address: [email protected] (J. Schloesser). 0257-8972/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.surfcoat.2011.07.084
temperatures and the liquid hydrogen (LH2) cooling inside the cooling channels, high thermal gradients are imposed and heat fluxes are expected to be in the range of 160 MW/m 2 [5]. To protect the material from the oxidizing environment and from high thermal loads, a protective coating could be applied. Although the thermo mechanical loading conditions differ, the main requirements of a thermal barrier coating (thermal and oxidation protection) are the same as in other applications. While first investigations were already published in the 80s (e.g. by Quentmeyer et al. [6]), several coating systems have also been considered in recent research. A thermal barrier coating system consisting of a NiCrAlY top coat and a Cu\Cr bond coat on Cu–8%Cr4%Nb-substrate has been investigated by Jain et al. [7] and Raj et al. [2,4]. These coatings were coldsprayed and hot-isostatically pressed and showed good adhesion in the initial thermal cycling tests. A TBC-system consisting of a MCrAlY-bond coat and zirconia was applied by electron beam vapor deposition by Schulz et al. [8]. This concept for TBC optimization on copper substrates starts from a well-known standard TBC system for nickel alloys. Although optimized for a nickel based substrate, a well-characterized coating system was chosen to investigate mainly the influence of the different loading conditions and the substrate. To enhance the lifetime of a suitable coating on copper substrates it is essential to know more about the failure mechanisms occurring in this new application. Unfortunately not much is published about the exact failure mechanisms.
1606
J. Schloesser et al. / Surface & Coatings Technology 206 (2011) 1605–1608
Due to the difference in materials' properties of copper-based (rocket engine) and nickel-based (turbines) alloys and to the difference in loading conditions, it is not clear whether the known failure mechanisms of thermal barrier coatings in gas turbines [9] can be transferred to rocket engines. Therefore, state-of-the-art thermal barrier coatings were applied to copper samples and exposed to thermal cycling and laser shock testing to investigate possible failure mechanisms. 2. Experimental procedures 2.1. Investigated materials Due to its good high temperature properties, the commercial alloy CuCrZr (specification number 2.1293), which is an age-hardenable copper alloy containing 1% chromium and 0.3% zirconium [10], was chosen as the substrate material. A standard coating system know from gas turbine applications was applied by atmospheric plasma spraying (APS) to the specimens. It consisted of a NiCrAlY bond coat (Ni–22%Cr–10%A–1%Y) and a partially stabilized zirconia (8% yttrium oxide) top coat. The coating system was chosen because it is already well known and characterized, although optimized for nickel based substrates. The coating parameters were optimized for the new substrate material [11], resulting in a well adhering coating.
been practiced in the past [12–15]. The advantage of this method is the well-defined heat flux and the contact-free heating compared to usual burner-rig-testing (an overview is given in [16]), that is commonly used to assess the lifetime of TBCs. The laser shock testing reported in literature was mainly performed with a moving laser spot (e.g. [15]). To realize a more homogeneous temperature distribution, a new set-up with a broad focal point was used in the present study (see Fig. 1). This is especially important as the copper substrate has a good heat conductivity. The diode laser used has a maximum power of 3 kW and the whole specimen surface can be heated homogenously. The surface temperature is controlled by a high speed pyrometer. Surface temperatures of up to 1500 °C can be reached within 0.2 s with the laser shock set-up while the backside of the samples is cooled by air. A surface temperature versus time plot is shown in Fig. 2. The plot also shows that the surface temperature is kept stable during the heating time of 2 s. Thin disk-shaped specimens with a thickness of 2 mm and a diameter of 20 mm were coated for laser shock testing with bond coat and thermal barrier coating. Zirconia has a poor absorption for the wavelengths used [17]. To ensure good absorption, the TBC was treated with a Fe3O4 powder suspension before testing. 3. Results and discussion
2.2. Thermal cycling experiments
3.1. Failure produced by thermal cycling
Thermal cycling experiments were performed on flat specimens with 35 mm width, 30 mm height and 5 mm thickness. One face of the specimens was completely coated (30 mm × 35 mm) while the other faces remained uncoated. The cycling experiments were performed to assess the adherence of the produced coatings and to examine the failure mechanisms. The specimens were heated up to 800 °C inside an oven, and, after a holding time of 30 min, they were cooled down in different cooling agents outside the oven. This cycle was repeated at least 50 times on each specimen in case of no failure of the coating system. The described cycling method is “isothermal” in the sense that all parts of the coating and the substrate are at the same temperature at the same time and are cooled and heated more or less simultaneously. No thermal gradient was applied to the coating. The temperature was also limited by the copper substrate. Nevertheless, due to the difference in the coefficients of thermal expansion, large thermal strains are imposed during heating and cooling of the specimen, which could lead to a spallation of the coating. The specimens were optically inspected after each cycle and after each 5–10 cycles one specimen was taken from the experiment for metallographic preparation. The prepared samples were investigated by optical microscopy as well as by scanning electron microscopy (SEM). To examine the effect of cooling rates, three different cooling agents were used: air, icy water and liquid nitrogen (LN2). The cooling rates were measured with a calibration specimen equipped with thermocouples. Air cooling shows the slowest cooling rate with about 0.5 K/s from 800 °C to room temperature. The fastest cooling rate, 13 K/s from 800 °C to 10 °C, is reached with icy water. Liquid nitrogen shows an in-between value with 7 K/s from 800 °C to −140 °C. Cooling with liquid nitrogen was applied because it reaches a low temperature (liquid hydrogen is used in real application), although due to gas formation at the surface of the specimen the cooling rate is not as fast as in icy water.
Specimens that were cooled down slowly (air cooling) did not show any failure of the thermal barrier coating after 50 cycles. A micrograph of a sample after exposure can be seen in Fig. 3. At the uncoated edges the copper oxidizes severely but the coating itself provides a certain oxidation protection. Due to the oxidation of the copper at the uncoated edges, an overhang of the coating was formed (Fig. 3). No spallation or partial delamination of the thermal barrier coating was observed with this cooling rate. At higher cooling rates (icy water and LN2) some of the specimens behaved similar to those cooled down slowly and did not show any sign of failure. Other samples showed partial delamination at the interface between substrate and bond coat (compare Fig. 4). A layered structure of copper oxide formed at these delaminated regions. Delamination started only at the uncoated edges where the copper oxide undermines the coating. A layer of copper oxide was formed in 1 cycle and then spalled during cooling, the next layer of copper oxide was then formed at high temperatures and the whole process was repeated. Raj et al. [2] observed a similar behavior in their oxidation tests at the uncoated sample edges, but in contrast to our results, the coatings were not undermined. GRCop-84 was in used in their study which has a better general oxidation resistance. It is believed that the substrate alloy plays a more important role concerning the susceptibility to this failure mode than the coating system. Statistical variations of the coatings are believed to be the reason why this phenomenon is seen only in part of the samples. This failure
2.3. Laser shock testing To account for the high thermal gradients in real service, a different testing method than thermal cycling is needed. Heating the samples on only one side by a laser is an approach that has already
Fig. 1. Set-up for laser shock testing.
J. Schloesser et al. / Surface & Coatings Technology 206 (2011) 1605–1608
1750
Surface temperature Set temperature
1500
Temperature [° C]
1607
1250 1000 750 500 250 0 0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
Time [s] Fig. 2. Surface temperature versus time for laser shock experiments showing high heating rates.
mechanism is critical under operational conditions because a totally dense coating cannot be assured due to erosion. This mechanism is mainly driven by oxidation of the substrate, and none of the coatings failed completely.
3.2. Failure produced by laser shock testing Laser cycling experiments have been performed at different surface temperatures from 750 °C and 1500 °C. A metallographic post mortem analysis was applied to each specimen. The specimens failed completely after the first cycle (2 s holding time at high temperatures) at temperatures above 1250 °C. With lower temperature, the samples survive more cycles. A failure plot for 2 s holding time is shown in Fig. 5. These results show that the thermal gradient plays an important role for the lifetime of the coatings as the thermal gradient is increasing with increasing surface temperature. Comparing the results to thermal cycling, the samples survived only 8 cycles at 750 °C compared with more than 50 cycles in thermal cycling experiments. This indicates clearly that it is not enough to test the coating systems with thermal cycling experiments without thermal gradients. Fig. 6 shows a typical failure after laser cycling. The thermal shock induced by the laser leads to a delamination of the whole coating at the substrate interface. In none of the experiments partial failure was observed. No sign of oxidation at the interface between substrate and bond coat was visible with scanning electron microscopy (SEM).
Fig. 4. Micrograph after 50 cycles (800 °C, LN2 cooling) showing partial delamination.
No formation of thermally grown oxide on the bond coat/substrate interface was observed. This was the case because the heating times were very short compared to normal thermal cycling test. But as the service times in rocket engines are also quite short it is not likely that a thermally grown oxide will form, which plays a crucial role in the failure mechanisms of thermal barrier coatings on nickel based substrates. Due to the large difference in thermal expansion coefficients of the bond coat and the substrate, large thermal stresses were probably induced in this region. These results show on the one hand that, again, the substrate interface is the weak interface and on the other hand that the laser heating method is a suitable method to qualify different coating systems because it produces damage and even delamination by thermal stresses. Therefore, it is an important tool for future coating development.
4. Conclusions Possible failure modes of TBCs on copper substrates were investigated in this paper. The observed failure modes differ significantly compared to specimens with nickel substrate (e.g. for gas turbine applications). Thermal cycling led to partial delamination due to copper oxide formation only at specimen edges for high cooling rates. The copper oxide was undermining the coating at the interface between bond coat and substrate. This failure mode is critical in case of a damaged layer e.g. due to erosion in a rocket engine. No complete spallation of the coatings occurred during thermal cycling. No formation of a thermally
2s holding time
Temperature [° C]
1500
1250
1000
750 no failure failure
500 0
2
4
6
8
10
Number of cycles Fig. 3. Micrograph after 50 cycles (800 °C, air cooling) showing no delamination.
Fig. 5. Cycles to failure depending on surface temperature (laser heating time 2 s).
1608
J. Schloesser et al. / Surface & Coatings Technology 206 (2011) 1605–1608
References
Fig. 6. Typical failure after laser cycling at the interface between substrate and bond coat.
grown oxide, which is critical in failure of thermal barrier coatings on gas turbines, was observed. To account for the thermal gradient inside the coating and fast heating rates, a laser shock experiment was set up to qualify the coatings. These experiments show delamination at the interface between substrate and bond coat. This failure mechanism will be investigated in subsequent studies. All experiments show that the interface between bond coat and the substrate is the critical interface for the investigated material system. For further thermal barrier coating development, the optimization of this interface is the main focus.
[1] J. Riccius, O. Haidn, E. Zametaev, 40th AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, 2004. [2] S.V. Raj, C. Barrett, J. Karthikeyan, R. Garlick, Surf. Coat. Technol. 201 (2007) 7222. [3] L.U. Ogbuji, Surf. Coat. Technol. 197 (2005) 327. [4] S.V. Raj, L.J. Ghosn, C. Robinson, D. Humphrey, Mater. Sci. Eng. A 457 (2007) 300. [5] S. Beyer, D. Klemm, M. Bobeth, W. Pompe, O. Trommer, P. Gawlitza, Adv. Eng. Mater. 7 (2005) 54. [6] R.J. Quentmeyer, H.J. Kasper, J.M. Kazaroff, AIANSAE 14th, Joint Propulsion Conference, 1978. [7] P. Jain, S.V. Raj, K.J. Hemker, Acta Mater. 55 (2007) 5103. [8] U. Schulz, K. Fritscher, M. Peters, D. Greuel, O. Haidn, Sci. Technol. Adv. Mater. 6 (2005) 103. [9] J. Rösler, M. Bäker, M. Volgmann, Acta Mater. 49 (2001) 3659. [10] Deutsches Kupferinstitut, Kupferdatenblatt CuCr1Zr, Deutsches Kupferinstitut, 2005. [11] J. Schloesser, T. Fedorova, M. Bäker, J. Rösler, J. Solid Mech. Mater. Eng. 4 (2010) 189. [12] W. Pompe, H.-A. Bahr, I. Pflugbeil, G. Kirchhoff, P. Langmeier, H.-J. Weiss, Mater. Sci. Eng. A 233 (1997) 167. [13] S. Rangaraj, K. Kokini, Acta Mater. 52 (2004) 455. [14] A.S. Semenov, H.-A. Bahr, H. Balke, H.-J. Weiss, Thin Solid Films 471 (2005) 200. [15] D. Nies, R. Pulz, S. Glaubitz, M. Finn, B. Rehmer, B. Skrotzki, Adv. Eng. Mater. 12 (2010) 1224. [16] R. Vaßen, F. Cernuschi, G. Rizzi, A. Scrivani, N. Markocsan, L. Östergren, A. Kloosterman, R. Mevrel, J. Feist, J. Nicholls, Adv. Eng. Mater. 10 (2008) 907. [17] J. Eldridge, C. Spuckler, J. Am. Ceram. Soc. 91 (2008) 1603.