Materials Science and Engineering A277 (2000) 64 – 76 www.elsevier.com/locate/msea

Effects of residual stress on the performance of plasma sprayed functionally graded ZrO2/NiCoCrAlY coatings K.A. Khor *, Y.W. Gu Materials Laboratory, School of Mechanical and Production Engineering, Nanyang Technological Uni6ersity, Nanyang A6enue, Singapore 639798, Singapore Received 16 February 1999; received in revised form 5 August 1999

Abstract Functionally graded ZrO2/NiCoCrAlY coatings were produced by plasma spraying using pre-mixed and spheroidized powders as the feedstock. The microstructure, density, elastic modulus, thermal conductivity/diffusivity, microhardness and coefficient of thermal expansion were found to change gradually through the five-layer functionally graded coatings which was beneficial for the improvement of mechanical and thermal properties of the coatings. The residual stresses of the as-sprayed coatings with different graded layers and different thicknesses, as well as the changes of residual stresses during thermal cycling were simulated by finite element analysis (FEA). Results showed that residual stress was the lowest for the five-layer functionally graded coating compared to that of the duplex coating and three-layer coating with the same thickness, and the residual stresses increased with a decrease in coating thickness. For the coatings with the same thickness, the bond strength and thermal cycling resistance were found to increase with an increase in the number of graded layers which is due to the decrease in the residual thermal stresses. The bond strength of the five-layer functionally graded coating was about twice as high as that of the duplex coating and the number of thermal cycles of functionally graded coating was five times higher than that of the duplex coating. Results also showed that the bond strength decreased with an increase in the coating thickness. © 2000 Elsevier Science S.A. All rights reserved. Keywords: Functionally graded material (FGM); Plasma spray; Residual stress; Bond strength; Thermal cycling; ZrO2; Thermal barrier coating

1. Introduction The development of thermal barrier coatings (TBCs) has led to highly efficient thermal barrier systems for gas turbines and diesel engines. Plasma spraying is an attractive and simple method of forming TBCs. However, conventional TBCs and direct ceramic/metal joining usually suffer premature failure because of spallation during cooling due to excessive residual stresses generated near the interface and poor bond strength between the coating and the substrate. Cracking and spallation usually occur at the substrate/coating interface or metal bond coat/ceramic top coat interface [1,2]. The residual stresses have a substantial effect on the coating properties [3]. They can give rise to deformation of coated workpieces and spallation (or cracking) of the coating. In addition, the nature of the * Corresponding author. Tel.: +65-7995526; fax: +65-7911859. E-mail address: [email protected] (K.A. Khor)

residual stresses significantly influences various types of coating properties, such as bond strength [4], thermal cycling [5,6] and erosion resistance [7], etc. In order to improve the long-term performance of TBCs, it is important and necessary to simulate the residual stresses and understand the coating failure mechanisms. In an effort to release the residual stress and improve the properties of plasma sprayed coatings, functionally graded material (FGM) with a graded composition from the top coat to the bond coat were designed in order to reduce thermal expansion mismatch among the different coating layers and substrate [8,9]. However, up to now, there are few investigations on the effectiveness of FGM coatings in the reduction of thermal stresses and improvement of the properties, and as we know, the poor bond strength between the coating and substrate of plasma sprayed TBCs is always a problem when these coatings are subjected to mechanical and thermal stresses.

0921-5093/00/$ - see front matter © 2000 Elsevier Science S.A. All rights reserved. PII: S 0 9 2 1 - 5 0 9 3 ( 9 9 ) 0 0 5 6 5 - 1

K.A. Khor, Y.W. Gu / Materials Science and Engineering A277 (2000) 64–76 Table 1 Plasma spraying parameters Primary gas (pressure); flowrate Auxiliary gas (pressure); flowrate Powder feed rate Arc current Arc voltage Spray distance

Argon (50 psi); 82 scfh Helium (50 psi); 26 scfh 30 g/min 800 A 50 V 120 mm

In this paper, ZrO2/NiCoCrAlY functionally graded coatings were prepared by plasma spraying using prealloyed and plasma spheroidized composite powders. The microstructure, microhardness, density, elastic modulus, thermal conductivity/diffusivity and coefficient of thermal expansion were studied. Coatings with different thicknesses and coating layers were also prepared for bond strength and thermal cycling resistance tests. Numerical simulation on residual stresses existed in these as-sprayed coatings and the stresses generated during thermal cycling were simulated using the ANSYS 53 finite element analysis code [10]. The results were used to correlate with the bond strength and thermal cycling resistance of the different types of coatings.

2. Experimental procedures The 8 wt.% Y2O3 stabilized ZrO2 and NiCoCrAlY powders (Praxair, USA) were used for preparing the FGM coatings. The mixture of these two types of powders with different ratios of NiCoCrAlY (25, 50 and 75% by weight) was ball milled and subsequently spheroidized by plasma spraying into distilled water using a 40-kW plasma torch (SG-100, Miller Thermal, USA). Table 1 shows the parameters used in the plasma spraying process. An optical microscope (Leica DMR, Switzerland) was used to study the microstructure of coatings. The microhardness (Vickers) profiles of duplex and FGM coatings were measured on cross-section specimens using a Matsuzawa DMH-1 microhardness tester (Japan), with a load of 300 g and duration time of 15 s. Elastic modulus of five individual layers of FGM coatings

65

under room temperature and high temperatures was measured using a four-point bending test with a high temperature 8502 Instron machine. The single-layer peel-off coatings with a dimension of 45× 8×0.8 mm3 were tested under 25, 400, 800 and 1200°C, respectively. The CTE of freestanding five individual layers was measured at a heating rate of 10°C/min as a function of temperature from 25 to 1000°C using a Perkin Elmer TMA 7 thermomechanical analyzer. Nickel stubs were used as the substrates for bond strength tests and the peel-off 5-mm thick duplex and five-layer FGM coatings were used in the thermal cycling test. Prior to plasma spraying, the substrate surfaces were grit blasted using silicon carbide grits followed by ultrasonic cleaning. The composition and thickness of different layers of duplex coatings and FGM coatings used in bond strength tests are shown in Table 2. The ASTM standard C633-79, i.e. the bond strength testing method for the plasma sprayed coatings, was used to measure the bond strength of coatings. Two identical cylindrical nickel stubs were used as a set, one with the coating on the surface and the other without. A high performance DP-460 Epoxy Adhesive (USA) with a maximum bond strength of 40 MPa was used to join the two stubs. The surface of the uncoated stub was sand blasted to enhance the adhesion strength. The two stubs were aligned and a weight of  420 g was applied to ensure an intimate contact between the two surfaces. After 12 h of curing at room temperature, the bond strength (the maximum stress at which two stubs are separated) was measured using an Instron 4302 tester at a cross-head speed of 1 mm min − 1. The results were the average of five samples. A high temperature Netzsch DIL 402 C dilatometer (Germany) which offers a maximum temperature of 2400°C with a sensitivity of 1 digit/1.25 nm was used to measure the thermal cycling resistance of both the duplex and FGM coatings. The 5-mm thick peel-off (duplex and FGM) coatings were heated and cooled in a vacuum oven cyclically between room temperature and 1300°C with an equal heating and cooling rate of 50 K/min. The sharp drop or increase in shrinkage or dilatation was considered as the thermal cycling failure of the coating.

Table 2 Coating layers and thicknesses of duplex and FGM coatings Number

100% N (mm)

25% N+75% Za (mm)

50% N+50% Z (mm)

75% N+25% Z (mm)

100% Z (mm)

Overall thickness (mm)

1 2 3 4

200 200 100 200

– – 100 200

– 200 100 200

– – 100 200

400 200 200 200

600 600 600 1000

(two-layer) (three-layer) (five-layer) (five-layer) a

N, NiCoCrAlY; Z, ZrO2.

K.A. Khor, Y.W. Gu / Materials Science and Engineering A277 (2000) 64–76

66

3. FEA considerations The residual stress distribution of duplex and FGM coatings was analyzed by the finite element method (FEM, ANSYS 5.3). The calculations were performed on a UNIX workstation. The thermal-structural element PLANE 13 was selected. For the modeling of residual stress after plasma spraying, the model used (shown in Fig. 1) represents a cylinder shape nickel substrate of 50 mm in diameter and 10 mm thick with a coating deposited to the top surface. The coating and substrate are assumed to be isotropic for simplicity in this study. The analytical model is a perfect elastic body without plastic deformation. An axial symmetric problem is chosen to reduce computer costs and data manipulation time. Three types of NiCoCrAlY/ZrO2 coatings (two-layer, threelayer and five-layer) with the thicknesses of 600 mm and 1 mm are computed. A fine mesh is introduced to model both the coating and the substrate. The elements and nodes of the different types of coating systems are shown in Table 3. Sand blasting treatment usually introduces compressive stresses on the substrate that are subsequently reduced owing to the temperature rise during spraying [11]. The high stresses presented in the deposition of powders exist for a very short time when the molten particles impinge on the substrate or on a previously deposited coated surface and the greater part of this stress is relieved by plastic flow at a much reduced yield stress [12]. For this reason, the quenching stress is relatively small, and, the entire specimen is assumed to be stress free at the temperature of 427°C (at which the spraying process is assumed to end) [13]. Thus, the residual stresses of as-sprayed coatings are considered to be caused by cooling from a uniform steady-state temperature of 427°C to room temperature (25°C).

In the thermal cycling experiments, 5-mm thick peeloff coatings with the length and width of 10 mm were used. The analysis method simplifies to a two-dimensional axisymmetric model when the symmetry of the configuration is considered. Both the models for FGM coatings (1-mm thick for each of five layers) and duplex coatings (2 mm for bond coat and 3 mm for ceramic top coat) consist of 625 four-node isoparametric solid elements and 676 nodes. The coating and the substrate are assumed to be isotropic and elastic for simplicity. The heat transfer boundary conditions used for the analysis are shown in Fig. 2. Compared with the upper surface cooling, the lower surface and edge surface heat transfer cooling are ignored, because only the upper surface (ceramic surface) is cooled with a cooling fan, and subjected to a convective heat transfer with a heat transfer coefficient of 1000 W/m2K [13]. The modeling consists of two steps. First, a thermal model is used to determine the temperature through the specimen during cooling. Second, the resulting thermal histories are transferred to a mechanical model to compute the thermal stresses both in the coating and in the substrate. Thermal radiation is not considered in the calculation. The in-plane thermal stresses are calculated through constructing a continuous model. The steps of thermal loading conditions are shown as follows: 1. Calculation of residual stresses owing to a uniform cooling from an assumed uniform stress-free manufacturing temperature of 427°C to room temperature; 2. A steady state analysis is performed to calculate the thermal residual stresses as the specimen is heated to 1300°C; 3. A cooling fan is applied on the surface of the coating via a heat transfer coefficient of 1000 W/ m2K which results in a different transient temperature distribution during cooling. From this, the

Fig. 1. Schematic description of the geometry used in the finite element model. Table 3 Elements and nodes for different kinds of coating systems

Element Node

Two-layer, 600 mm

Five-layer, 600 mm

Two-layer, 1000 mm

Three-layer, 1000 mm

Five-layer, 1000 mm

1300 1377

1500 1581

1300 1377

1500 1581

1500 1581

K.A. Khor, Y.W. Gu / Materials Science and Engineering A277 (2000) 64–76

Fig. 2. Schematic diagram of boundary condition for thermal cycling.

transient thermal stresses owing to surface cooling of the coating are obtained. Elastic modulus, density, coefficient of thermal expansion, thermal conductivity and specific heat for different coating layers at different temperatures were experimentally measured for the computation [14]. The thermal diffusivity/conductivity of different layers was obtained by laser flash method. The results of these properties are shown in Table 4. Poisson’s ratios of pure Ni based alloy and ZrO2, together with the property of Ni substrate were obtained from Ref. [15]. The Poisson’s ratios of interlayers, composed of the ceramic top coat and metal bond coat without considering porosity, can be assumed from the following equation [2], n0 = nmVm + nc (1−Vm )

(1)

in which n0, nc and nm are the Poisson’s ratio of interlayer, ceramic coating and metal coating, respectively. Vm is the volumetric ratio of metal. The effect of the change of Poisson’s ratio on the residual stress is little, so Poisson’s ratio can be expressed as a linear function of the position. The effects of porosity will not be considered in this finite element analysis.

Fig. 3. CTE of individual layers of FGM coating.

4. Results and discussions

4.1. Characterization of FGM coatings Fig. 3 shows the coefficient of thermal expansion (CTE) of individual layers of NiCoCrAlY/ZrO2 FGM coating as a function of temperature from 25 to 1000°C. As can be observed from Fig. 3, CTE increases with an increase in temperature and changes gradually through the five individual layers of FGM coating. There is a significant difference in the CTE value between ZrO2 layer and NiCoCrAlY layer and the difference increases significantly with an increase in

Fig. 4. Elastic modulus of individual layers of FGM coating.

67

b

a

25 400 800 25 400 800 1200 25 400 800 1200 25 400 800 1200 25 400 800 1200 25 400 800 1200

From Ref. [15]. From Ref. [10].

100% ZrO2

75% ZrO2

50% ZrO2

25% ZrO2

NiCoCrAlY

Ni substrate

Temperature (°C)

Density (kg/m3) 8880 8880 8880 7320 7320 7320 7320 6960 6960 6960 6960 6626 6626 6626 6626 6208 6208 6208 6208 6037 6037 6037 6037

Elastic modulus (Pa) 207×109 182×109 150×109 225×109 186×109 147×109 90×109 187×109 193×109 141×109 76×109 158×109 146×109 89×109 61×109 105×109 119×109 93×109 52×109 53×109 52×109 46×109 48×109

Table 4 Material properties of five individual layers [14]

1.27×10−5a 1.64×10−5a – 1.4×10−5 2.4×10−5 4.7×10−5 7.1×10−5 1.2×10−5 1.91×10−5 3.65×10−5 6.4×10−5 1.1×10−5 1.9×10−5 3.5×10−5 5.2×10−5 9.11×10−6 1.51×10−5 2.96×10−5 3.96×10−5 7.2×10−6 9.4×10−6 1.6×10−5 2.2×10−6

Coefficient of thermal expansion

0.312b 0.312 0.312 0.3 0.3 0.3 0.3 0.2875 0.2875 0.2875 0.2875 0.275 0.275 0.275 0.275 0.2625 0.2625 0.2625 0.2625 0.25b 0.25 0.25 0.25

Poisson’s ratio

90.5 65.3 73.9 4.3 6.4 10.2 16.1 3.3 3.5 3.7 6.2 3.1 3.8 5.6 8.5 2.7 3.0 3.9 5.1 1.5 1.2 1.2 1.1

Thermal conductivity (W/mK)

461 460 460 501 592 781 764 535 672 725 743 517 621 689 719 519 629 716 734 500 576 637 656

Specific heat (J/kg °C)

68 K.A. Khor, Y.W. Gu / Materials Science and Engineering A277 (2000) 64–76

K.A. Khor, Y.W. Gu / Materials Science and Engineering A277 (2000) 64–76

69

Fig. 5. Microhardness distribution of duplex and FGM coatings.

Fig. 6. Cross-section microstructure of (a) 5-layer FGM coating and (b) duplex coatings.

temperature. Fig. 4 shows the variation of elastic modulus of five individual layers of FGM coating with an increase in temperature. It can be observed that the elastic modulus of the different layers decreases accordingly with an increase in temperature. There is a dramatic decrease in the elastic modulus of NiCoCrAlY layer with an increase in temperature whereas there is little change in elastic modulus with an increase in temperature for ZrO2 layer. The elastic modulus decreases with an increase in the content of ZrO2. From the microhardness distributions of FGM coating and duplex coating shown in Fig. 5, it can also be observed that the microhardness changes gradually through the five-layer FGM coating whereas a significant microhardness difference exists between NiCoCrAlY and ZrO2 layers for the duplex coating. Fig. 6(a) and (b) shows the cross-section micrographs of FGM coating and duplex coating, respectively.

From Fig. 6(a), it can be observed that NiCoCrAlY gradually changes its distribution pattern from lamellar pattern to dispersed pattern from NiCoCrAlY layer to ZrO2 layer and the ZrO2 layer also changes gradually. No clear interface between the two adjacent layers can be observed. The gradient distribution of the two phases in the coating can significantly decrease the high thermal stress generated due to the sharp differences in coefficient of thermal expansion and elastic modulus between the two phases, NiCoCrAlY and ZrO2. For duplex coating as shown in Fig. 6(b), there is a distinct boundary between porous ZrO2 layer and dense NiCoCrAlY layer. Because of the large differences in elastic modulus, coefficient of thermal expansion and hardness between the two layers for duplex coating, there will be large thermal and mechanical stresses generated at the interface of the two layers during the thermal and mechanical loading. It is these large thermal stresses

70

K.A. Khor, Y.W. Gu / Materials Science and Engineering A277 (2000) 64–76

that cause the deterioration of bond strength and spallation or cracking of the duplex coating.

4.2. Residual stress distribution in as-sprayed FGM and duplex coatings 4.2.1. Fi6e-layer FGM coating Fig. 7 shows the distribution of the radial stress on the surface and interfaces of the coating plotted along the radius of the coating after cooling from the spraying temperature of 427°C to room temperature. It can be observed that the surface is usually compressive and the compressive stresses change to tensile stresses gradually with an increase in the distance from the surface into the interface of the coating/substrate [16,17]. At the surface and interfaces of 0% Ni/25% Ni, 25% Ni/50% Ni and 50% Ni/75% Ni, the radial stresses remain compressive. The compressive stresses decrease abruptly near the edge of the specimen. At the interface of coating and substrate and the interface of 100% NiCoCrAlY layer and 75% NiCoCrAlY layer, the stresses are tensile, and the maximum tensile stress of 40 MPa is near the edge of the specimen. The large radial stresses on the surface of the coating may cause the formation of surface crack. Fig. 8(a) shows a typical contour plot of axial stress distribution for five-layer coating. Large tensile stress is generated near the edge of the specimen and at the interface of the specimen and it will cause the spallation of the coating [18]. This tensile stress decreases significantly with the increase of distance from the edge of the specimen. There is a compressive stress inside the coating near to this large tensile stress as shown in Fig. 8(a).

Fig. 8(b) shows the distribution of axial stresses in the interfaces. The maximum tensile axial stress of 49 MPa between the interface is obtained at the edge of the specimen. The axial stresses tend to decrease quickly with the increase in the distance from the edge. This large stress concentration near the interface and the edge of the specimen can cause the spallation of the coating. Fig. 9(a) shows the contour plot of shear stress distribution for five-layer FGM coating. The large interface shear stresses are related to the interface crack. The maximum stress (− 14 MPa) is obtained near the edge within the 50% NiCoCrAlY layer of the coating, as can be seen from Fig. 9(a). With the increase in the distance from the edge to the center, the stresses decrease. The shear stress distribution at the interfaces along radius is shown in Fig. 9(b). The shear stresses show a remarkable stress concentration at or close to the edge of the specimen. The stress concentration near the edge may cause the spallation of the coating.

4.2.2. Effect of the number of layers for FGM and duplex coatings with the same thickness Fig. 10(a)–(c) shows the maximum radial stress, axial stress and shear stress in 1-mm thick two-layer duplex coating along with three-layer and five-layer FGM coatings. As can be observed from Fig. 10, the highest residual stresses are obtained in the duplex coating. The residual stresses produced after plasma spraying processing can be relaxed with the application of threelayer and five-layer FGM coatings. For the two types of FGM coatings, the residual stresses generated in five-layer FGM coating are relatively lower in compari-

Fig. 7. Radial stress distribution along radius for five-layer coating.

K.A. Khor, Y.W. Gu / Materials Science and Engineering A277 (2000) 64–76

71

Fig. 8. Axial stress distribution of five-layer coating.

son with those of the three-layer FGM coating, because of a lower thermal expansion mismatch. The maximum radial stress for five-layer FGM coating decreases by 8 MPa compared with that of the duplex coating, while

the maximum axial stress and shear stress decrease by 23 and 16 MPa, respectively. The maximum radial stress of all three types of coatings is on the surface of the coating. The maximum axial stress is above the

72

K.A. Khor, Y.W. Gu / Materials Science and Engineering A277 (2000) 64–76

bond coat and shear stress is below the ZrO2 layer which may generate the interface crack. From the above analysis, the five-layer FGM coating has the lowest stresses compared with those of the

duplex coating and three-layer FGM coatings. The radial stress, axial stress and shear stress results all show the minimum value for the five-layer FGM coating.

Fig. 9. Shear stress distribution of five-layer coating.

K.A. Khor, Y.W. Gu / Materials Science and Engineering A277 (2000) 64–76

73

600-mm thick FGM coating compared with 1-mm thick coating decrease by 6, 14 and 7 MPa, respectively. The effect of the thickness on the stresses in the duplex coatings is almost similar to that of the FGM coatings. The thicker the coating, the larger the residual stresses in the coating [19].

4.3. Bond strength

Fig. 10. Comparison of stresses in different numbers of layers with same thickness.

4.2.3. Effect of coating thickness For the five-layer FGM coatings with a thickness of 600 mm and 1 mm, the distributions of residual stresses are almost the same but the stress concentration becomes more severe for the 1-mm thick coating. The maximum radial stress, axial stress and shear stress for

The bond strength of FGM coatings (with different layers but the same thickness) prepared by manual plasma spraying is shown in Fig. 11. Five-layer NiCoCrAlY/ZrO2 FGM coating has the highest bond strength (13 MPa), while duplex coating exhibits the lowest bond strength (5.46 MPa). With an increase in number of intermediate layers, the bond strength of the coating with the same thickness increases. Results from finite element analysis shown in Fig. 10 can be used to explain these experimental results. As can be observed from Fig. 10, the radial stress, axial stress and shear stress are extremely high in duplex coatings. It is these high stresses which cause the spallation, cracking and delamination of the duplex coating. The residual stresses produced after the plasma spraying process can be significantly relaxed with the application of three-layer and five-layer FGM coatings as can be found from Fig. 10. The residual stresses generated in five-layer FGM coating are relatively lower in comparison with three-layer FGM coating, which can be used to explain the relatively high bond strength of five-layer FGM coatings. Bond strength is observed to decrease with an increase in the coating thickness as shown in Table 5. One reason is that the larger the coating thickness, the larger the residual thermal stresses, and this will cause the degradation of bond strength. Compared to the radial stresses, axial stresses, as well as the shear stresses of the 600-mm thick five-layer FGM coatings, the stress concentration for 1-mm thick coating is more severe. For duplex coating, it is also revealed that the thicker the coating, the larger the residual stresses in the coating. However, compared with an FGM coating with the same thickness, the duplex coating reveals a much higher residual stress value. Another reason is that with an increase in the coating thickness, the defects, such as interplanar voids, pores, cracks, unmelted particles and incomplete bonding between lamellae will be increased, and these are detrimental to the adhesion and cohesion of the as-sprayed coating.

4.4. Residual stress and thermal cycling resistance 4.4.1. Finite element analysis results The finite element analysis of thermal cycling is concentrated on the simulated thermal stresses during a cooling period of one thermal cycle from room temper-

74

K.A. Khor, Y.W. Gu / Materials Science and Engineering A277 (2000) 64–76

Fig. 11. Bond strength of coatings with different layers (coating thickness: 600 mm).

ature to 1300°C. The distributions and changes of the radial stress, axial stress and shear stress during this cooling period for FGM and duplex coatings were compared. Fig. 12(a) shows the maximum radial stress distribution (in the surface of ZrO2 layer) during cooling for duplex coating and FGM coating at different times. The duplex coating has larger radial stresses than FGM coating, which indicates that the duplex coating is more prone to surface cracking than FGM coating during thermal cycling. The radial stress for FGM coatings reaches the maximum value of 127 MPa after 5 s, then decreases with the increase of the cooling duration. This phenomenon is because during thermal cycling, when the cool air shocks the coating surface, the surface coating shrinks significantly to cause the transient maximum tensile stress generated at the surface of the coating. This transient maximum tensile stress will cause the generation of the orthogonal cracks on the coating surface [20]. Orthogonal cracks may open in the ceramic top coat and grow due to tensile stresses during cooling [21,22]. With the increase in the cooling duration, the differences in temperature for both duplex coating and FGM coating decrease and therefore, the thermal stresses gradually decrease. Fig. 12(b) and (c) shows the maximum tensile axial stress and tensile shear stress during cooling for the duplex and FGM coatings. As expected, the FGM coating has lower axial stresses and shear stresses, and this phenomenon indicates that the duplex coating is easy to spall during the cooling process of thermal cycling. The maximum tensile axial and shear stresses are obtained after 10 s of cooling period for both the

FGM coating and duplex coating. This high transient tensile stress will cause the generation of interface cracks and cause the spallation of coatings. With an increase in cooling duration, the residual stresses gradually decrease, as shown in Fig. 12(b) and (c).

4.4.2. Thermal cycling resistance Duplex and FGM 5-mm thick coatings were heated and cooled cyclically between room temperature and 1300°C. The sharp increase in dilatation was considered as the failure mechanism when the coatings failed under thermal cycling. The thermal cycling results of duplex coating and five-layer FGM coating are shown in Table 6. It can be observed that the resistance of the FGM coating to the thermal cycling is much better than that of the duplex coating. The performance of plasma sprayed coatings during thermal cycling is related directly to the generation of thermal fatigue cracks due to the role of thermal residual stress in coatings. According to the finite element analysis results shown in Fig. 12, the duplex coating has higher radial stress than the FGM coating which indicates that duplex coating is more prone to surface cracking than FGM coating during thermal cycling. Also the lower axial stresses and shear stresses in the Table 5 Bond strength of five-layer FGM coating with different thicknesses Thickness (mm)

Bond strength (MPa)

0.6 1

13.000 6.979

K.A. Khor, Y.W. Gu / Materials Science and Engineering A277 (2000) 64–76

75

the significant decrease in residual stresses and the increase in cohesive strength.

5. Conclusions The microstructure, microhardness, elastic modulus, coefficient of thermal expansion and thermal conductivity/diffusivity were found to change gradually in the five-layer FGM coating. For 1-mm thick five-layer as-sprayed FGM coating, the maximum compressive radial stress (− 72 MPa) is found to be at or near the surface of the specimen where surface cracking may be generated. The maximum axial stress (49 MPa) is at the edge of the specimen where spallation may occur. The maximum shear stress (− 14 MPa) is also at or close to the edge. With an increase in the number of layers with the same thickness of coatings, the bond strength of coatings increases. The bond strength of as-sprayed FGM coating is twice as high as that of the duplex coating because of the significant reduction in the residual thermal stresses. The duplex coating is prone to cracking and delamination due to the mismatch of thermal expansion coefficient between the ZrO2 and NiCoCrAlY layers. It can be verified that the stress concentration in the coating is reduced by using the functionally graded distribution of the coating. The five-layer graded coating is a feasible design because its architecture can prevent the crack initiation of the coating. With an increase in the coating thickness, the bond strength decreases, which can be attributed to the higher residual stresses and an increase in the defects in the thicker coating. During thermal cycling, the surface cracks and the interface cracks of the coating are usually produced by the transient tensile stress which is the largest at the onset of the cooling process. The thermal cycling life of five-layer FGM coating is five times higher than that of the duplex coating which can be indicated from the FEA results.

Acknowledgements

Fig. 12. Stress distribution during cooling for duplex and FGM coatings.

The authors would like to thank the support of the School of Mechanical and Production Engineering, Nanyang Technological University, Singapore in the form of research grants RP 56/92 and RG 25/96. Table 6 Thermal cycling resistance of duplex and FGM coatings

FGM coating indicate that the duplex coating is easy to spall during the thermal cycling cooling process. In brief, the reason for the improvement of thermal cycling resistance for FGM coating can be attributed to

Thermal cycling (cycles)

Duplex

Five-layer FGM

15

90

76

K.A. Khor, Y.W. Gu / Materials Science and Engineering A277 (2000) 64–76

References [1] R.A. Miller, C.E. Lowell, Thin Solid Films 95 (1982) 265 – 273. [2] Y.G. Jung, S.C. Choi, C.S. Oh, U.G. Paik, J. Mat. Sci. 32 (1997) 3841 – 3850. [3] R. Kingswell, K.T. Scott, D.T. Gawne, in: E. Broszeit, W.D. Munz, H. Oechsner, K.T. Rie, G.K. Wolf (Eds.), Plasma Surface Engineering, Deutsche Gesellschaft fur Metallkunde, Oberursel, FRG, 1989, pp. 695–700. [4] Y. Yang, Z. Liu, C. Luo, Y. Chuang, Surface Coatings Technol. 89 (1997) 97 – 100. [5] R. Hamacha, B. Dionnet, A. Grimaud, F. Nardou, Surface Coatings Technol. 80 (1996) 295–302. [6] D.W. Jordan, K.T. Faber, Thin Solid Films 235 (1993) 137 – 141. [7] R. Elsing, O. Knotek, U. Balting, Surface Coatings Technol. 43/44 (1990) 416–425. [8] Y.W. Gu, K.A. Khor, Y.Q. Fu, Y. Wang, Surface Coatings Technol. 96 (1997) 305–312. [9] K.A. Khor, Y.W. Gu, Z.L. Dong, in: Proceedings of the 15th International Thermal Spray Conference, Nice, France, May, 25 – 29, 1998, pp. 1543–1549. [10] G.J. DeSalvo, R.W. Gorman, ANSYS, Engineering Analysis System, User’s Manual, vol. I/II, Swanson Analysis System, Houston, TX, 1989.

.

[11] M. Ferrari, J.H. Harding, M. Marchese, Surface Coatings Technol. 48 (1991) 147 – 154. [12] S. Kuroda, T. Fukushima, S. Kitahara, Thin Solid Films 164 (1988) 157 – 163. [13] K. Kokini, Y.R. Takeuchi, B.D. Choules, Surface Coatings Technol. 82 (1996) 77 – 82. [14] Y.W. Gu, Ph.D. Thesis, Nanyang Technological University, Singapore, 1999. [15] J. Shackelford, W. Alexander, The CRC Materials Science and Engineering Handbook, CRC Press, Boca Raton, FL, 1992. [16] S. Takeuchi, M. Ito, K. Takeda, Surface Coatings Technol. 43/44 (1990) 426 – 435. [17] P. Bengtsson, C. Persson, Surface Coatings Technol. 92 (1997) 78 – 86. [18] M. Nishida, T. Hanabusa, H. Fujiwara, Surface Coatings Technol. 61 (1993) 47 – 51. [19] N. Perrin, H. Burlet, M. Boussuge, C. Desplanches, Surface Coatings Technol. 56 (1993) 151 – 156. [20] J. Musil, M. Alaya, R. Oberacker, J. Thermal Spray Technol. 6 (4) (1997) 449 – 455. [21] P.M. Pierz, Surface Coatings Technol. 61 (1993) 60–66. [22] K. Kokini, B.D. Choules, Y.R. Takeuchi, J. Thermal Spray Technol. 6 (1) (1997) 43 – 49.