Effects of deposition temperature and thermal cycling on residual stress state in zirconia-based thermal barrier coatings

Effects of deposition temperature and thermal cycling on residual stress state in zirconia-based thermal barrier coatings

Surface and Coatings Technology 120–121 (1999) 103–111 www.elsevier.nl/locate/surfcoat Effects of deposition temperature and thermal cycling on resid...

379KB Sizes 104 Downloads 149 Views

Surface and Coatings Technology 120–121 (1999) 103–111 www.elsevier.nl/locate/surfcoat

Effects of deposition temperature and thermal cycling on residual stress state in zirconia-based thermal barrier coatings V. Teixeira a, *, M. Andritschky a, W. Fischer b, H.P. Buchkremer b, D. Sto¨ver b a University of Minho, IMAT-Institute of Materials — Physics Department, P-4700 Braga-Portugal b Forschungszentrum Ju¨lich GmbH, IWV-Institute for Materials and Processes in Energy Systems, D-52425 Ju¨lich, Germany

Abstract Advanced ceramic multilayered coatings are commonly used as protective coatings for engine metal components to improve performance, e.g. thermal barrier coatings ( TBCs). Zirconia-based TBCs were produced by plasma spraying process and characterized in terms of microstructure, porosity, elastic modulus, adherence and residual stresses. In this contribution the residual stresses in multilayered coatings applied on Ni based superalloys for use as thermal barrier coatings were studied both by numerical modelling and experimental stress measurement. The thermal residual stresses generated during the spraying process of duplex TBCs were simulated by using an heat transfer finite element program and an elasto-plastic biaxial stress model. The TBC system was subjected to different thermal cycling conditions (maximum temperature, heating up and cooling down rates, dwell time at maximum temperature, etc.). The stress distribution within the TBC was also modelled after thermal cycling. The stress state in the as-deposited and in thermally cycled coatings was verified using an X-ray diffraction technique. The measurements were in good agreement with the residual stress modelled calculations. It was observed that the residual stresses were dependent on the thermal history of the TBC (as-deposited and thermally cycled). It is proposed that thermal cycling allowed the stresses to relax by microcracking and creep mechanisms at high temperature such that on cooling down to room temperature, an in-plane biaxial compressive stress will arise on the zirconia top coating due to the difference on the coefficients of thermal expansion between the metallic substrate and ceramic coating material. © 1999 Published by Elsevier Science S.A. All rights reserved. Keywords: Plasma spraying of zirconia; Residual stress analysis; Stress and failure during thermal cycling; Thermal barrier coatings

1. Introduction Ceramic thermal barrier coatings ( TBCs) have been recognized as showing considerable promise for a variety of applications involving high heat flux and/or moderately high temperature environments such as gas turbines [1,2]. Demands for increased efficiency and performance of gas turbine engines are met in part by increasing combustion temperatures and reducing cooling systems. Temperature increases within the engine approach the service limits of current superalloys. TBC systems on superalloys combine a thermal insulating zirconia top coat (providing a lower temperature in the metallic substrate), which has a relatively low thermal conductivity and large coefficient of thermal expansion compared with other technological ceramic, with a met* Corresponding author. Tel.: +351-53-604334; fax: +351-53-678981. E-mail address: [email protected] ( V. Teixeira)

allic bond coat that protects the substrate from both oxidation (usually with the development of an alumina protective thin scale) and hot corrosion when operated at high temperature. The structure of a TBC systems consists, traditionally, of a thick ceramic top coating (stabilized zirconia) deposited by atmospheric plasma spraying (APS ) on Ni superalloys precoated with a metallic bond coating (typically MCrAlY where M is Ni or/and Co) applied by vacuum plasma spraying ( VPS). Alternatively, this duplex coating system can be prepared by PVD processes (physical vapour deposition) such as electron beam evaporation ( EB-PVD) [1,2]. Another coating concept, which is also analysed in this contribution, is to substitute the metallic VPS bond layer with a thinner, dense ceramic coating (ZrO 2 coating stabilized by Y O or a composite coating 2 3 ZrO –Y O –Al O ) deposited by d.c. reactive magnetron 2 2 3 2 3 sputtering (PVD bond coating) [3]. The idea of using a thin PVD bond coating resides on the fact that the coating microstructures achieved by PVD processes are

0257-8972/99/$ – see front matter © 1999 Published by Elsevier Science S.A. All rights reserved. PII: S0 2 5 7- 8 9 7 2 ( 9 9 ) 0 0 34 1 - 2

104

V. Teixeira et al. / Surface and Coatings Technology 120–121 (1999) 103–111

dense and columnar, so these structures may act as an efficient diffusion gas barrier and control the oxidation of a metallic substrate. The thermal strain tolerance and thus the thermal shock resistance are improved due to the segmented nature of PVD coatings. Furthermore, a compressive coating pre-stress, induced by the sputtering process, balances the tensile thermal stress at high temperature [3,4]. Residual stresses in TBCs play an important role in the performance and lifetime of the coated component. Residual stresses within the coating system are generated by both the coating deposition process and service. One source of intrinsic residual stress within the as-deposited TBC is the thermal residual stress of plasma-sprayed zirconia coating (PS top coating), which arises from the nature of the plasma-spraying process associated with the rapid cooling of molten droplets, impacting on the cool substrate. Once rapid solidification is complete, the constraint of the underlying substrate (or previously solidified material ) restrains any further thermal contraction in the lamella, and a tensile microstress distribution (‘quenching stress’) is established because the temperature of substrate remains relatively constant. Stress relaxation by through thickness microcracking in the lamellar structure of the splat occurs as the ceramic material is unable to support high tensile stresses [5]. The other source of thermal stress occurs during cooling down to room temperature after the coating deposition. Due to the different coefficients of thermal expansion (CTE) and/or temperature gradients, a thermal residual stress develops in the ceramic–metal coating system, which can cause delamination at the interface or fracture of the material [6,7]. Away from the edges, the in-plane stresses (parallel to the interface) are typically compressive within the ceramic (since it has a lower CTE) and tensile in the metallic substrate. The final stress state will be a result of the combined quenching stress and cool down stress. Cracking in the ceramic coating or interfacial decohesion will affect the thermo-mechanical integrity of the coated component. TBC failure is expected to occur either from transient thermal stresses during rapid thermal cycling or from compressive stresses developed on cooling after a high-temperature exposure [4,6–9]. In the later case, the failure is observed on oxide scales that were grown at high temperatures in a stress-freestate. A compressive in-plane stress can cause cracking parallel to, or at, the coating/substrate interface, leading to spalling or delamination failures. The mechanisms for microcrack propagation at interfaces in coatings with compressive stresses have been analysed previously [5,6 ]. Under compressive stress, spallation may result either from the growth of a tensile, wedge crack along the interface or from buckling and cracking of the oxide coating. The mechanism of spallation of ceramic coatings under lateral compression is determined by the

relative fracture strengths of the ceramic–metal interface. Two processes can arise: wedging or buckling. Wedging arises when the strength of the interface is higher than the compressive fracture strength of the ceramic layer. By contrast, a relatively weak interface leads to a buckling process. The unavoidable interfacial oxidation process when TBC is at a high temperature in air will contribute to TBC failure. First, new interfaces appear, with a thin oxide layer growing together with new associated stress fields (intrinsic stress induced by growth and thermal stresses due to the CTE mismatch). Second, the density of microdefects in oxide layers is another factor to take into account, either by the presence of microporosity, by the initial growth, or later on, of a mixture of oxides such as spinel Ni(Cr,Al ) O , which are very brittle and 2 4 have a lower adherence compared to a-Al O and can 2 3 induce TBC failure by the combined effect of stress and the decrease in adherence [3,9–12]. In this contribution, we analyse the coating residual stress experimentally and numerically in the TBCs for the as-deposited coatings and after different thermal cycling conditions. Both residual stress measurement and modelling contribute to a better understanding of the failure mechanisms of the plasma-sprayed thermal barrier coating system during operation at high temperature.

2. Experimental 2.1. TBC preparation The TBC consisted of a top coating and a bond coating attached to a nickel-based substrate material, Inconel 617. As top coating powder material, 7.52 wt% yttria partially stabilized zirconia with an agglomerated and sintered morphology was chosen. The size range for the ZrO –8 wt% Y O was 10–45 mm. The metallic 2 2 3 bond coating material was a gas-atomized NiCoCrAlY powder. The chemical composition was Ni22Co17 Cr12.5Al.6Y with a size range of 22.5–45 mm. A 300 mm ZrO –8 wt% Y O thick top coat was deposited on top 2 2 3 of the bond coat (previously deposited by low-pressure plasma spraying) in a Sulzer Metco-F4 plasma spraying unit. The deposition of the zirconia overlayers was performed, typically at a gun power of about 48 kW, at atmospheric pressure (APS coatings). The zirconia top coating was also deposited using the inert high-presssure plasma spraying mode (IPS coatings) [12]. Table 1 shows the main deposition parameters for the preparation of the TBC systems studied. Fig. 1 schematically shows the two types of TBCs studied. The deposition parameters for the zirconia PVD bond layers prepared by d.c. reactive magnetron sputtering are listed in Table 2. Details on the conditions of deposition of

V. Teixeira et al. / Surface and Coatings Technology 120–121 (1999) 103–111

105

Table 1 Plasma-spraying parameters used to deposit the metallic bond coating and the zirconia top layer Parameter

APS layera

IPS layerb

VPS layerc

Spraying mode Coating type Gun power (kW ) Current (A) Plasma gas: Ar/H /He ( l min−1) 2 Carrier gas Ar ( l min−1) Nozzle diameter (mm) Powder flow rate (g min−1) Plasma gun-substrate distance (mm) Chamber pressure (mbar) Substrate temperatured ( K ) Spraying efficiency (%) Coating thickness (mm)

Atmospheric ZrO –8 wt%Y O 2 2 3 48 650 30/15/0 3 7 19.9 150 1000 500/700/900 54 300

Inert gas ZrO –8 wt%Y O 2 2 3 48 590 30/15/0 3 7 19.9 150 1990 500/700/900 80 300

Vacuum NiCoCrAlY 47 750 20/8/25 2 7 – 250 100 750 – 150

a Atmospheric plasma spraying. b High-pressure inert gas plasma spraying. c Vacuum plasma spraying. d The substrate temperature during spraying was 500, 700 or 900 K, depending on the experiment.

zirconia PVD layer, PS top coatings and protective behaviour during high-temperature exposure of duplex TBCs can be found in Refs. [3,10,12]. 2.2. High-temperature tests The TBCs were tested by isothermal cycling in a furnace and rapid thermal cycling with a natural gas– oxygen torch [8]. The isothermal cycling consists of a 10 min heating period from 200°C to a maximum temperature of either 1000 or 1100°C with a 45 min holding time, followed by a cooling period for 10 min with static air. The rapid thermal cycling with the natural gas– oxygen flame involved heating up to either 1000 or 1100°C within 1 min and then cooling down to about 200°C within 1 min by forced air. The temperature was measured by a thermocouple clamped into a bore hole within the metallic substrate. After the high-temperature tests, the samples were examined by optical microscopy, scanning electron microscopy, energy dispersive X-ray

spectroscopy, Micro-Raman spectroscopy and X-ray diffraction analysis ( XRD). 2.3. Residual stress analysis — experimental The in-plane surface residual strains/stresses in the coatings were determined by X-ray diffraction using the sin2 y technique (see Fig. 2). When a biaxial stress exists in the irradiated layers of the surface coating, the lattice strains are related to the stress by [13]: 1+n n d −d wy 0= s sin2 y− (s +s ), (1) w 22 d E E 11 0 where d is the unstressed lattice spacing, d is the 0 wy stressed lattice spacing, s is the stress component along w the direction w defined in the plane of the coating, y is the tilt angle, E is Young’s modulus and n is the Poisson ratio. Eq. (1) predicts a linear relationship between d wy and sin2 y, and the stress can be obtained from a least-

Fig. 1. Schematic of the cross section of an Inconel substrate coated with: (i) a duplex PVD/PS thermal barrier coating and (ii) a traditional thermal barrier coating with the VPS metallic bond coating. The typical layer thicknesses are also represented.

106

V. Teixeira et al. / Surface and Coatings Technology 120–121 (1999) 103–111 Table 2 Deposition parameters for the d.c. reactive magnetron sputtering Parameter

PVD coating

Area of the Zr target (m2) Area of the Y (m2) Y O content (wt%) 2 3 ZrO content (wt%) 2 Ar pressure (Pa) O pressure (Pa) 2 Sputter current (A) Deposition rate (m s−1) Coating thickness (m) Substrate temperature (°C ) Substrate bias ( V )

5.4×10−3 1.9×10−4 11 Bal. 0.9 0.1 4.4 5.5×10−10 8.75×10−6 280 −40 (r.f.)

Fig. 2. Coordinate system used in residual stress determination.

squares fit of experimentally determined d-spacings, measured at a number of y tilts (see Fig. 3). The zirconia [620] lattice plane was used to obtain the coating surface strain. The corresponding diffraction angle occurred at about 2h=144.5°. The y angle was varied from −45 to +45° using 13 steps to determine the corresponding inter-planar distance d . wy A full Pseudo-Voigt profile fit of two reflections with background correction was done as well as a crosscorrelation after corrections with respect to absorption, background, Lorentz and polarization [13]. XRD data were collected for each coating before and after thermal cycling. It should be mentioned that the XRD strain/stress measurements are limited to the near-surface regions of the coating due to the low X-ray penetration in zirconia (typically limited to several micrometres in depth, depending on the porosity). Therefore, the results from the XRD were compared to the stress modelled at the surface position of the top coating, and

Fig. 3. Typical d-spacing of the tetragonal [620] crystallographic plane as function of different y-tilts of the TBC sample (measurement at coating surface). Two measurements are shown: one for an as-deposited TBC and one after thermal cycling. The slopes indicate an increase in compressive stress after the thermal cycling process.

it was assumed that any changes in the interfacial stresses would be reflected in the measured surface residual stress [14–16 ]. The residual stress was measured after removal of the surface roughness typical for the plasma-sprayed coatings. The stress measured by XRD in the as-deposited original rough surface would be affected by a local stress relaxation within this rough surface, which has dimensions comparable to the penetration depth for the X-rays [14,15]. The roughness was removed by a careful polishing with a very small load using 0.25 mm diamond suspension as the last polishing step. The effect on the residual stress state of this polishing procedure was observed as a small increase towards the compression state when comparing the surface coating stress before and after the polishing (generally less than 20% of the stress value measured on the original rough surface). 2.4. Residual stress analysis — modelling The simulation of residual stresses developed during the plasma-spraying coating deposition is based on a finite element program to solve the unsteady heat conduction equation. With the computed temperature gradients in incremental time steps of the plasma spraying process, a biaxial elasto-plastic thermal stress model is used to calculate the stress distribution within the plasma-sprayed coating, using temperature-dependent physical properties. To predict the thermal residual stress profile during and after thermal cycling of the TBC system, a two-part numerical model was used. First, the temperature profile through the multilayered structure was numerically calculated from the solution of the onedimensional heat transfer equation with constant boundary conditions using a finite difference code (an implicit approach, the Crank–Nicolson method, was chosen because it provides a second-order accuracy in both space and time). Second, the coating thermal residual stress was computed from the temperature history of the sample using isotropic and temperature-dependent physical properties (see Table 3) from data reported in

107

V. Teixeira et al. / Surface and Coatings Technology 120–121 (1999) 103–111 Table 3 Physical properties (at room temperature) for the modelled materials Physical property

Ni alloy

NiCoCrAlY

Al O 2 3

Cr O 2 3

ZrO Y O 2 2 3

Density (kg m−3) Thermal Conductivity ( W m−1 K−1) Heat Capacity (J kg−1 K−1) Emissivity Young’s modulus (GPa) Poisson ratio Thermal expansion coefficient (10−6 K−1) Yield stress Rp0.2% (MPa)

8230 13.4 419 0.5 214 0.31 12 360

8100 12.5 400 0.5 170 0.25 12.6 300

3970 3.6 779 0.7 380 0.25 5.4 300

5210 1.2 815 0.75 265 0.29 8.68 300

5700 1.8 450 0.8 70 0.23 8.6 138

the literature [17–21]. The stress in each element of the coating/substrate system is obtained by solving the n independent equations resulting from the equilibrium condition, the strain compatibility at each interface and the equilibrium of bending moments [15,22], as represented in the following equations: n ∑ F =0 i i=1 e =e i i+1 n n ∑ M +∑ F i i i=1 i=1

(2) (3)

A

B

1 i ∑ t = t =0. j 2 i j=1

(4)

3. Results and discussion 3.1. Temperature during plasma-spraying deposition Fig. 4 shows the measured substrate temperature and the simulated temperature at the same location in the Inconel substrate (1 mm from the uncoated surface) during the spraying of the top coating. The experiment involved pre-heating the substrate to 950 K, followed

Fig. 4. Measured and modelled substrate temperature evolution during the deposition of the yttria stabilized zirconia top coating.

by ten passages of the plasma gun for the spraying of the zirconia top coating at 700 K. At the end of the coating deposition, the coating system was cooled down to room temperature. The simulation using temperaturedependent physical properties (see Table 3) shows a good agreement with the measured temperature profile. From the temperature measurements and simulations, it can be seen that the general trend is the same during spraying of the metallic bond coating and the ceramic top coating, and an increase in substrate temperature occurs during the first pre-heat passages of the plasma gun. During the spraying process, there is a rapid temperature decrease for the individual powder particles, from the melting temperature to the temperature of the underlying material. Simultaneously, the temperature of the underlying material is increased due to the heat conduction. The mean substrate temperature slowly decreases as the insulating zirconia layer increases in thickness (Fig. 4). This is experimentally observed and is predicted by our numerical model. 3.2. Residual stresses in the as-deposited coatings The residual stress for the as-deposited coatings was measured by X-ray diffraction at room temperature, and the results show that the level of stress at the top coating surface was very low for both the traditional and duplex PVD-PS thermal barrier coatings. The stress changed from about 40 to −240 MPa with increasing substrate temperature during deposition from 500 to 900 K ), which was in agreement with the results of stress modelling (see Fig. 5). Tensile stresses within the ceramic coating, deposited on the substrate kept at 500 K, indicate a greater degree of shrinkage of the coating as compared with the substrate. The opposite effect, leading to compressive residual stresses within the coating, is observed at 900 K. The experimentally determined residual stresses were found to be lower than the calculated residual stresses. However, from Fig. 5, the correlation is acceptable, taking into account several modelling assumptions [8,17] and some degreee of uncertainty in the zirconia plasmasprayed coating physical properties, such as the in-plane

108

V. Teixeira et al. / Surface and Coatings Technology 120–121 (1999) 103–111

processes, we measured the residual stress in the samples after each manufacturing process step. For the duplex PVD-PS thermal barrier coatings, the residual stresses within the PVD bond coating were measured by XRD, micro-Raman spectroscopy and laser microdisplacement (substrate curvature). The PVD bond coating has an in-plane compressive residual stress after the deposition (about −300 MPa), which is mainly due to the atomistic deposition process (see discussion below and Table 4). 3.3. Residual stresses in the thermally cycled coatings

Fig. 5. Comparison between the measured and the modelled results on residual stress at the top coating surface for several TBC systems deposited at different substrate temperatures. Zirconia APS and zirconia IPS means plasma-sprayed zirconia (by atmospheric and inert gas technique, respectively) deposited on Inconel 617 pre-coated with the PVD coatings. NiCoCrAlY/zirconia is the conventional TBC deposited onto Inconel 617.

Young’s modulus and the thermal conductivity [17,18]. In fact, modelling of thermal stresses requires knowledge of material elastic properties such as stress–strain data ranging from room temperature to a very high temperature. The modelling results show that the as-deposited coatings have a linear stress gradient with a higher compressive stress at the interface, decreasing to very low levels at the surface, in agreement with earlier work [7,16,23–25]. Since it is important to develop a fundamental understanding of the stress generation during all steps of the coating deposition and thermal cycling

The residual stresses after thermal cycling were determined for the two types of TBCs. In the case of the duplex PVD-PS thermal barrier coating, the stress was measured for all the layers including the oxide scale grown during the high-temperature testing. During thermal cycling, it is important to analyse the coating stress within the three layers: the growing oxide interlayer, the bond layer and the plasma-sprayed top coat. In the case of duplex PVD-PS coatings, the Cr O layer grows between the PVD layer and the 2 3 metallic substrate (Inconel 617 which is a chromia forming alloy), by oxygen diffusion through the PVD layer [3,12]. We assume that the volume increase due to the oxidation of the chromium is relatively low, and therefore Cr O layer is stress-free at the growth temper2 3 ature. During the cooling cycle, a compressive stress develops within the Cr O layer due to the CTE mis2 3 match of coating and substrate [15]. The residual stresses within the chromia layer obtained (−1680 MPa by Micro-Raman spectroscopy and −2060 MPa by XRD analysis) are in good agreement with the numerical model results (approximately −1800 MPa) [14,15]. For

Table 4 Residual stress in individual layers in the as-deposited and thermally cycled state of Duplex PVD/PS thermal barrier coatings as determined by several methods State of the layer

Layer typea

Model calculation (MPa)

X-ray diffraction (MPa)

Raman spectroscopy (MPa)

Curvature laser transducer (MPa)

As-deposited As-deposited After thermal cycling at 1000°C As-deposited As-deposited As-deposited As-deposited After isothermal cycling at 1100°C After isothermal cycling at 1000°C After rapid thermal cycling at 1000°C After thermal cycling After thermal cycling

PVD layer PVD-P layer PVD layer APS-5 APS-7 APS-9 IPS-9 APS-5 IPS-9 APS-7 Cr O 2 3 Cr O c 2 3

– – −640 52 12 −5 −5 −280 −230 −120 −1800 –

−280 −1200 −800 38 10 −17 −25 −240 −108 −71 −2060 −1100 to −2500

−320 −1300 −760 – – – – – – – −1680 –

−270 −1580 b – – – – – – – – – –

a The code PVD-P means that the substrate was polished with SiC-P1200; otherwise, it was grit-blasted APS-5, APS-7. IPS-9, etc. means the mode of deposition of the ZrO –8 wt% Y O top coat (e.g. APS-5 is an atmospheric plasma-sprayed zirconia coating deposited at 500 K, IPS-9 is 2 2 3 a high-pressure plasma spraying coating deposited at 900 K, etc.). b Thin zirconia PVD coating measured on a thin glass substrate. c Values from Refs. [26,27] reported as chromia scales grown on NiCr and Inconel alloys.

109

V. Teixeira et al. / Surface and Coatings Technology 120–121 (1999) 103–111

Table 5 Residual stress in the as-deposited and thermally cycled state of conventional NiCoCrAlY/zirconia plasma-sprayed thermal barrier coatings as determined by numerical modelling and the X-ray diffraction technique State of the layer

Layer type

Model calculation (MPa)

X-ray diffraction (MPa)

As-deposited As-deposited As-deposited As-deposited As-deposited After isothermal cycling at 1100°C After rapid thermal cycling at 900°C After rapid thermal cycling at 900°C As-deposited As-deposited After thermal cycling After thermal cycling After thermal cycling

APS-5 top coat APS-7 top coat APS-9 top coat IPS-7 top coat NiCoCrAlY APS-7 top coat APS-5 top coat IPS-7 top coat TBC-lit.b TBC-lit.c TBC-lit.b TBC-lit.d Al O scale 2 3

52 12 −5 12 50 −280 −100 −100 10–20 −40 to 40 – – −2800

31 10 −12 4 80–120a −110 −62 −73 10–35 −10–25 −150 −290 −2500 to −3300e

a Result from Ref. [7] measured by the layer-by-layer removal method. b Value from Ref. [16 ] reported to NiCoCrAlY/ZrO –7 wt% Y O on Al–12% Si piston heads. 2 2 3 c Value from Ref. [23] reported to NiCrAlY/ZrO –7 wt% Y O on Ni alloy. 2 2 3 d Value from Ref. [30] reported to NiCoCrAlY/ZrO –8 wt% Y O on Inconel 600. 2 2 3 e XRD results from Refs. [28,29] reported as alumina scales grown on NiCr alloys.

the alumina layer grown on top of the NiCoCrAlY bond coat, the numerical model gave a higher stress level of about −2800 MPa, which is consistent with the different elastic properties of alumina scales. After thermal cycling, the PVD bond coating showed an increase in compressive stress, probably due to stress relaxation during the high-temperature exposure (see Tables 4 and 5 and Fig. 6). When the TBC system is cooled down to room temperature after a high-temperature exposure (without thermal cycling), the PS zirconia coating develops a compressive stress of about −120 MPa, lower than the

measured values for the as-deposited coatings and not as high as for the furnace cycled coatings. Low values of about −70 to −120 MPa were determined experimentally for both traditional and duplex PVD/PS TBCs subjected to a low number of cycles or short periods at high temperature. The surface zirconia stress increased after prolonged heat treatments or after a large number of thermal cycles at high temperature (−183 to −240 MPa), which is in agreement with the model calculations (about −280 MPa, see Fig. 6). Let us consider the stress calculation of the plasmasprayed top coating. The zirconia top coating is in a

Fig. 6. Residual stress evolution for conventional TBCs and duplex PVD/PS thermal barrier coatings as a function of the thermal treatment.

110

V. Teixeira et al. / Surface and Coatings Technology 120–121 (1999) 103–111

tensile stress state at working temperatures of 1100°C. During heat treatment, it is assumed that the coating relaxes almost immediately, by microcracking, to a tensile stress level of about 140 MPa, and during prolonged heat treatment, the thermal stress is relaxed by creep within the metallic substrate and eventually within the coating [8,10,31]. Stress relief by oxide creep is likely to be much less significant because of the relatively low rate of oxide creep at these high temperatures as compared with metal creep rates since both the metallic substrate and zirconia coating are held at a maximum temperature for a short period [31–33]. A tensile stress in the zirconia coating will cause through-thickness cracks [10,34]. The coating stress relief mode at high temperature is believed to be cracking of the brittle plasma-sprayed zirconia top coating. Microcracking for ceramic coatings undergoing thermal cycling was detected by acoustic methods by other authors [35]. Due to this stress relaxation at high temperature, a compressive thermal stress is then again generated within the coating during the cooling procedure (see Fig. 6, Tables 4 and 5). After the heat treatment, the top coating develops on cooling a higher compressive stress due to the relaxation processes at high temperature. The stress model gives −280 MPa within the upper layer of the PS top coating, which agrees well with the XRD measurement of a TBC tested for a long period at 1100°C without any visible damage (about −240 MPa). As can be seen in Fig. 6, there is a general tendency for an increase in compressive residual stress after furnace cycling for the duplex PVD/PS TBCs (and also observed for the conventional TBCs). An annealed TBC, generally, presents a higher value of in-plane compressive stress (see Tables 4 and 5) possibly due to a lower coating damage induced by the thermal cycling process (occurrence of microcracking and buckling at the interfaces).

4. Conclusion A numerical model predicting the thermal residual stresses within plasma-sprayed TBCs was applied to evaluate the residual stresses generated in the zirconia top coating after deposition. The model proposes that a stress gradient exists through the coating thickness, i.e. the stresses change gradually to an in-plane compressive stress towards the interface. The residual stress in the as-deposited top coating changed from tensile to compressive with increasing substrate temperature. The correlation between modelled results and XRD measurements was found to be acceptable, both revealing the same tendency with the variation of substrate temperature during deposition and coating thicknesses. It was shown that the residual stresses are dependent on the thermal history of the thermal barrier coating.

Thermal cycling allows the stresses to be relaxed at high temperatures (this is believed to be due to microcracking of the ceramic coating and creep mechanisms of the metallic bond coating) such that on cooling down to room temperature, an in-plane biaxial compressive stress will arise due to the difference in CTE between the substrate and ceramic coating. The measurements for thermally cycled TBCs were in good agreement with the residual stresses calculated using a thermal stress model for multilayered systems subjected to thermal cycling processes. No significant differences were found for the residual stress in the zirconia top coating (as-deposited and after thermal cycling) in either the traditional TBC or the duplex PVD-PS TBC.

References [1] P. Vincenzini, Industr. Ceram. 10 (1990) 113. [2] J.T. DeMasi-Marcin, D.K. Gupta, Surf. Coat. Technol. 68–69 (1994) 1. [3] M. Andritschky, V. Teixeira, L. Rebouta, H.P. Buchkremer, D. Sto¨ver, Surf. Coat. Technol. 76–77 (1995) 101. [4] V. Teixeira, M. Andritschky, H. Gruhn, W. Mallener, H.P. Buchkremer, D. Sto¨ver, in: C. Berndt (Ed.), Proc. 8th National Thermal Spray Conf., ASM, Houston, TX, 1995, p. 515. [5] S. Kuroda, T.W. Clyne, Thin Solid Films 200 (1991) 49. [6 ] A.G. Evans, J.W. Hutchinson, Int. J. Solids Struct. 20 (1984) 455. [7] M.D. Thouless, J. Vac. Sci. Technol. 9 (1991) 2510. [8] P. Bengtsson, T. Johannesson, J. Thermal Spray Technol. 4 (3) (1995) 245–251. [9] R.A. Miller, J. Am. Ceram. Soc. 67 (1984) 517. [10] V. Teixeira, M. Andritschky, H.P. Buchkremer, D. Sto¨ver, Failure Mechanisms in Thermal Cycled TBCs, J. Lecomte-Beckers, F. Schubert, P.J. Ennis ( Eds.), Materials for Advanced Power Engineering 1998, Proc. 6th Lie`ge COST Conf. Vol. 5-III, Publ. Forschungszentrum Ju¨lich- Central Library, Ju¨lich, 1998, pp. 1601–1610. [11] B.C. Wu, E. Chang, C.H. Chao, M.L. Tsai, J. Mater. Sci. 25 (1990) 1112–1119. [12] V. Teixeira, M. Andritschky, L. Rebouta, H.P. Buchkremer, D. Sto¨ver, in: Proc. 3rd European Congress on Thermal Plasma Processes,-TPP3, Neuschu¨tz, Aachen, 1995, p. 445. VDI Ber-1166. [13] I.C. Noyan, J.B. Cohen, Residual Stress: Measurement by Diffraction and Interpretation, Springer, Berlin, 1987. [14] V. Teixeira, M. Andritschky, W. Fischer, H.P. Buchkremer, D. Sto¨ver, T. Ericsson ( Ed.), Proc. 5th Int. Conf. Residual Stresses-ICRS’5, Linko¨ping, Sweden Vol. 1 (1997) 436–441. [15] V. Teixeira, M. Andritschky, W. Fischer, D. Sto¨ver, Y.M. Haddad (Ed.), Advanced Multilayered and Fibre-reinforced Composites — Problems and Prospect Vol. 3/43 (1998) 415–430. [16 ] P. Scardi, M. Leoni, L. Bertamini, M.M. Marchese, Surf. Coat. Technol. 86–87 (1996) 109–115. [17] H.J. Groß, W. Mallener, D. Sto¨ver, R. Vaßen, in: Proc. 5th National Thermal Spray Conf., ASM International, Anaheim, USA, 1993, p. 581. [18] D. Schneider, T. Schwarz, H.P. Buchkremer, D. Sto¨ver, Thin Solid Films 224 (1993) 177. [19] Thermophysical Properties of Matter, Y.S. Touloukian ( Ed.), TPRC Data SeriesIFI, Plenum, New York, 1970. [20] G. Samsonov, The Oxide Handbook, IFI-Plenum Data Company, New York, 1982.

V. Teixeira et al. / Surface and Coatings Technology 120–121 (1999) 103–111 [21] Metals Handbook, 9th edition. Vol. 3 ASM, Metals Park, OH, 1980.. [22] V. Teixeira, M. Andritschky, W. Fischer, D. Sto¨ver, H.P. Buchkremer, in: Proc. 10th National Thermal Spray Conf. NTSC’97, ASM International, Indianapolis, IN, 1997, p. 839. [23] M. Levit, I. Grimberg, B.Z. Weiss, Mater. Sci. Eng. A206 (1996) 30–38. [24] F. Kroupa, J. Thermal Spray Technol. 6 (3) (1997) 309–319. [25] P. Bengtsson, C. Persson, Surf. Coat. Technol. 92 (1997) 78. [26 ] U. Marewski, Forschungszentrum Ju¨lich, 52425 Ju¨lich, Germany, Report Ju¨l-2330, (1989). [27] C. Liu, A.M. Huntz, J. Lebrun, Mater. Sci. Eng. A160 (1993) 113. [28] A.M. Huntz, J.L. Lebrun, A. Boumaza, Oxid. Met. 33 (1990) 321. [29] K.L. Luthra, C.L. Briant, Oxid. Met. 26 (1986) 397.

111

[30] A.H. Bartlett, R.D. Maschio, J. Am. Ceram. Soc. 78 (4) (1995) 1018. [31] G. Kerkhoff, R. Vaßen, C. Funke, D. Sto¨ver, Materials for Advanced Power Engineering 1998, J. Lecomte-Beckers, F. Schubert, P.J. Ennis ( Eds.), Proc. 6th Lie`ge COST Conf. Vol. 5-III, Publ. Forschungszentrum Ju¨lich, Ju¨lich, 1998, pp. 1669–1677. [32] J.H. Stout et al., Mater. Sci. Eng. A120 (1989) 193. [33] A.M. Freborg, B.L. Ferguson, W.J. Brindley, G.J. Petrus, Mater. Sci. Eng. A245 (1998) 182. [34] K. Kokini, Y.R. Takeuchi, B. Choules, Surf. Coat. Technol. 82 (1996) 77. [35] C.C. Berndt, J. Mater. Sci. 24 (1989) 3511.