Acoustic emission evaluation of fracture characteristics in thermal barrier coatings under bending

Surface & Coatings Technology 232 (2013) 710–718

Contents lists available at ScienceDirect

Surface & Coatings Technology journal homepage: www.elsevier.com/locate/surfcoat

Acoustic emission evaluation of fracture characteristics in thermal barrier coatings under bending L. Yang ⁎, Z.C. Zhong, J. You, Q.M. Zhang, Y.C. Zhou ⁎, W.Z. Tang Key Laboratory of Low Dimensional Materials & Application Technology (Ministry of Education), Xiangtan University, Xiangtan, Hunan 411105, China Faculty of Materials, Optoelectronic & Physics, Xiangtan University, Xiangtan, Hunan 411105, China

a r t i c l e

i n f o

Article history: Received 9 November 2012 Accepted in revised form 20 June 2013 Available online 28 June 2013 Keywords: Thermal barrier coating Acoustic emission Three-point bending Fracture toughness Fracture characteristics

a b s t r a c t The real-time assessment on the details in damage evolution of thermal barrier coatings (TBCs) is desirable, especially if the key coating performance parameters, such as, the surface and interface fracture toughness, could be accurately characterized. In this paper, the fracture details of as-sprayed and pre-oxidized TBCs under three-point bending are monitored by an acoustic emission (AE) combined with digital image correlation (DIC) methods. The surface and interface toughness of TBCs can be accurately determined on the basis of AE signals and strain images. A linear relationship is found between the energy released from coating failure and that of AE signals, whose slope depends on the fracture modes and properties of TBCs. © 2013 Elsevier B.V. All rights reserved.

1. Introduction With their high-melting, low heat conductivity, excellent wear resistance and high hardness, thermal barrier coatings (TBCs) have been widely regarded as an attractive material in enhancing the hightemperature limits of gas turbines and internal combustion engines [1–4]. In general, TBCs consist of four layers: a yttria stabilized zirconia (YSZ) ceramic top coating (TC) that increases the operation temperature of turbine components, a nickel-based substrate that endures mechanical loading, a MCrAlY alloy (M represents Ni, Co, or Fe) bond coating (BC) that enhances adhesion of the ceramic coating to substrate and a thermally grown oxide (TGO) layer that formed between bond and top coats due to the diffusion and reaction of oxygen and metal iron during processing and further thermal exposure. Each layer in the multi-layer structure of TBCs has remarkably different physical, thermal and mechanical properties, which result in internal stresses in TBCs. Moreover, the protected components, such as, turbine blades, vanes, combustors, commonly have very complex shape and withstand external mechanical stresses and aggressive environments. The complex shape, structure, and harsh operating conditions lead to unpredictable coating failures with two main types: surface vertical cracking in ceramic coating and cracking along top/bond or TGO interface [5,6]. The resistance to surface or interface crack formation and propagation determines the performance and tolerance for mechanical or thermal loading, which finally governs the durability of TBCs. Therefore, the characterization of the mechanical parameters, such as, surface and ⁎ Corresponding authors at: Faculty of Materials, Optoelectronic & Physics, Xiangtan University, Xiangtan, Hunan 411105, China. Tel.: +86 731 58293586. E-mail addresses: [email protected] (L. Yang), [email protected] (Y.C. Zhou). 0257-8972/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.surfcoat.2013.06.085

interface fracture toughness, which can be used to evaluate cracking resistance in TBCs, has been an important subject of research [7–11]. Compared with the formation of surface or interface cracks, the propagation and coalescence of these cracks are more important to the durability of TBCs because they determine the occurrence time for eventual largescale cleavage or spallation of coating. Therefore, the assessment of details in surface and interface cracking is necessary to understand the failure mechanism and further predict the service life of TBCs, and become an actively pursuing research issue [12–15]. Various methods have been developed to characterize the surface or interface fracture toughness of TBCs, such as surface or cross section indentation test [7], barb pushout test based on shear stress [8], bending test [9], double cantilever [10] and blister test [11]. Fracture toughness can be evaluated based on the values of critical crack-driven load and corresponding crack length, which should be measured in the experiment. In fact, the initiation of cracking is very difficult to measure because of successive loading and rapid propagation, especially for brittle ceramic coating, which results in difficulty in ascertaining the critical crack-driven load [11]. Therefore, an in-situ microscopy observation, which could provide viable and direct information on the crack evolution, is necessary, and has obtained a growing interest in the development of characterization equipments. However, the in-situ test is of very limited use in the failure evaluation of TBCs due to its small testing area and high cost. Thus, combining a real-time, nondestructive testing technique with a traditional mechanical parameter characterization method is of great importance to track the initiation of a crack and its propagation in a TBC specimen. The initiation of cracks in a material produces acoustic emission (AE) signals due to the release of locally stored elastic energy. Therefore, the cracking time and corresponding value of critical crack-driven load can

L. Yang et al. / Surface & Coatings Technology 232 (2013) 710–718

be directly ascertained by the initiation of AE behavior. Thus, a combination of experiment under destructive loading with real-time nondestructive AE testing is desirable for characterizing the fracture toughness of TBCs. More importantly, with the advantage of loose experimental condition and integral failure testing in a material, AE has been widely applied in the real-time detection of damage in TBCs [16–20]. In our recent works, the surface and interface cracks have been discriminated by using the wavelet analysis and frequency spectra of AE signals [21,22]. A linear correlation between the surface crack density in TBCs and AE events under the cyclic heating has been established [23], based on the assumption of proportional correlation of energy released from cracking and the corresponding AE signals. However, this assumption lacks experimental verification, and the details in fracture behavior, such as the characteristics of evolution in surface and interface cracking, are still intractable. Contrary to tensile or thermal loading, tensile stress is not uniform in the TC layer along the long axis direction of TBCs under the threepoint bending. Thus, only a few surface cracks occur in the region which bears the maximum bending moment. To clearly present the process of surface and interface cracking and limit the number of cracks, a three-point bending test was conducted to characterize the fracture toughness and analyze the evolution feature of TBCs. In addition, the high temperature oxidation results in the growth of TGO at interfaces of TBC layers, and changes the microstructure of the TC layer due to ceramic sintering. Therefore, the high temperature oxidation should be considered because it is a key factor that affects the mechanical properties as well as the failure mechanism. In this paper, the characterization of fracture toughness, details of fracture behavior, correlation of cracking of TBCs and corresponding AE signals are systematically investigated by using a combination of threepoint bending test and AE testing. First, a theoretical analysis on fracture toughness and energy estimation of surface and interface cracking in TBCs is developed. Then, the AE response of TBCs and strain evolution are recorded to characterize fracture toughness and analyze the details of fracture behavior. The results obtained from the AE testing can be used to analyze the correlation between the damage evolution in TBCs and AE characteristics. Finally, the influence of high temperature oxidation on fracture mechanism is analyzed based on AE characteristics.

2. Theoretical analysis 2.1. Fracture toughness and fracture energy for surface cracking

h i. 2 2 ρ ¼ ðL=2Þ þ δ 2δ

ð2Þ

where L and δ are the half length of the bending distance and the bending deflection at the middle point of the beam, respectively. The distance between the coating and the neutral axis of the composite beam is y ¼ h0 þ hc

ð3Þ

where h0 is the distance between the top/bond coating interface and the neutral axis of the composite beam. According to the requirement of static equilibrium, the distance can be expressed as  . 2 2 ð2Es hs þ 2Ec hc Þ h0 ¼ Es hs −Ec hc

ð1Þ

ð4Þ

A vertical surface crack in the ceramic TC layer that is generated during the three-point bending test can be regarded as a model-I crack. Thus, given that the critical crack-driven stress σcr and the corresponding crack length a0 are determined, the surface toughness can be calculated by K ΙC ¼ Y Ι σ cr

pffiffiffiffiffiffiffiffi πa0 ;

GΙC ¼

K 2IC Ec

ð5Þ

where YΙ is a geometry factor, and equals to 1 in the case of bending test [25]. As a typical brittle material, the length of a surface crack can be presented by coating thickness. If the number of surface cracks N is obtained from a metallographic microscope, then the fracture energy EFS can be calculated as EFS ¼ 2hc bN

K 2IC Ec

ð6Þ

where b is the width of a TBCs specimen. Assuming that a proportional relationship exists between the AE energy and that released from cracking (fracture energy), the AE energy of surface cracks ESAE can be expressed by S

Ec hc þ Es hs , E and h are the Young's modulus and the thickhc þ hs ness of layers, subscripts c and s denote the TC layer and substrate, respectively. The initial residual stress σr, its measurement details have been described in our previous work [24], is − 57 MPa and − 200 MPa for as-received and pre-oxidized TC layers, respectively. y and ρ are the distance between coating and the neutral axis of where E ¼

composite beam, and the curvature radius of the neutral axis, as shown in Fig. 1. The curvature radius can be expressed as

EFS ¼ α s EAE

Cracks in TBCs are strongly dependent on the transient stress. Stresses in TBCs consist of external loading and initial residual stress. The former is caused by the thermal or mechanical loading. The latter originates from the rapid contraction of sprayed splat during cooling from the deposition temperature. Residual stresses in ceramic coatings induced by the preparation are compressive because of the small thermal expansion coefficient of ceramic compared to that of the alloy substrate. The BC layer can be simplified to the same layer as substrate because of their similar mechanical properties. Thus, given that TBCs are only subjected to the elastic deformation, stress in the TC layer under bending can be expressed as Ey σ¼ þ σr ρ

711

ð7Þ

where αs is a coefficient. 2.2. Fracture toughness and fracture energy for interface cracking Under bending or tension, the interface crack energy release rate and interface stress intensity factor of TBCs can be calculated by the Suo–Hutchinson model [26] " # 2 2 c1 P M PM G¼ þ þ 2 pffiffiffiffiffi 2 sin γ 16 Ahc Ih3c AIhc vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi # u 2" 2 up P M2 PM t þ þ 2 pffiffiffiffiffi 2 sin γ K¼ 2 Ahc Ih3c AIhc

ð8Þ

ð9Þ

where P and M are the equivalent tensile load and bending moment per unit width, respectively. Under the three point bending loading, they can be determined by the following equations [26]. P ¼ σ r hc þ

  Σ 1 1 P ðt Þl −Δ þ I0 η 2 4bh

ð10Þ

712

L. Yang et al. / Surface & Coatings Technology 232 (2013) 710–718

80 mm 40 mm

AE sensor Substrate

BC

TC

CCD camera

AE system

DIC system

Fig. 1. Schematic AE, DIC monitoring apparatus and specimen dimensions of the three-point bending test.

M¼

Σ P ðt Þl 12I 0 4b

ð11Þ

where P(t) is the applied load in the three-point bending test. In Eqs. (8) to (11), the physical meaning and expression of dimensionless parameters A, p, I, Σ, I0, Δ, η and γ can be found in reference [25]. Obviously, given that the critical crack-driven loading Pcr is determined, fracture toughness KIIC and GIIC that are used to assess the resistance against an interface crack can be calculated using Eqs. (8) and (9), respectively. For an interface crack with a half length of a, the released fracture energy EFI can be expressed as EFI ¼ 2abG ½P ðt a Þ

ranges of 45–75 μm was deposited on the substrate surface, and ZrO2-8 wt.%Y2O3 ceramic coating with particle size ranges of 80–115 μm was deposited on the free surface of the bond coating. Spraying distance and velocity for bond and top coat were set to 180 mm at 10 g/min and 100 mm at 5 g/min, respectively. The spraying voltage and current were chosen as 40 V and 450 A for these two layers. The thicknesses of the substrate, bond coating and ceramic top coating are 2 mm, 150 μm and 400 μm, respectively. The length of the coatings is 40 mm. Some specimens were pre-oxidized at 1000 °C for 300 h in air to investigate the influence of high-temperature oxidation on the failure mechanism of TBCs.

ð12Þ 3.2. Three-point bending test and AE testing

where P(ta) is the applied load. Then, the assumed proportional relationship between the AE energy for interface cracking EIAE and the energy released from interface cracking can be expressed as I

EFI ¼ α I EAE

ð13Þ

where αI is a coefficient. The discrimination of AE energy for a surface crack ESAE and an interface crack EIAE will be discussed in Section 4.3. 3. Experimental 3.1. TBC specimens Ni-based superalloy GH3030 with the dimensions of 80 × 9 × 2 mm3 was chosen as substrate. Using the air plasma spray method, bond coating which composition is NiCr22Al7Y0.2 with particle size

As shown in Fig. 1, three-point bending tests were performed with RG2000-10 Universal Testing machine at room temperature. A straight pin perpendicular to the long axis of specimens was used to transmit the load. Bending load is applied at the middle of the TBC specimen along its long axis with a span width of 60 mm. The movement of the upper fixture was controlled with a rate of 0.2 mm/min. During process of bending, the fracture of TBCs was monitored by an AE system (PCI-2). The threshold, pre-amplifier and sampling frequency was set to 38 dB, 40 dB and 2 MHz, respectively. Two AE sensors were placed on the specimen surface, as illustrated in Fig. 1. To avoid noise in experiments, AE system was set that only signals between two sensors are recorded. A digital image correlation (DIC) system (AIMAS) was applied to measure the evolution of strain in TBCs. Images were obtained at a sampling rate of one picture every 2 s. Then these images were analyzed through software (Aramis 3-D) to extract the strain field.

L. Yang et al. / Surface & Coatings Technology 232 (2013) 710–718

4. Results and discussion 4.1. Fracture details of TBCs under bending To characterize the fracture toughness and analyze the correlation of TBCs and AE signals of TBCs, the mechanical parameters including Young's modulus of the TBCs layers, critical loads for surface and interface cracks should be determined. To obtain these parameters and analyze the influence of high-temperature on TBCs performance, a finite element simulation of the load–displacement curve of TBCs under bending was performed with application of the successive approximation method. Once the simulated load–displacement curve is nearly coincident with the experimental one, as shown in Fig. 2, the mechanical parameters used in the simulation can be selected. The results of Young's modulus are about 60 GPa and 80 GPa for as-sprayed and pre-oxidized TC layers, respectively, which are approximate to the reported values [27]. The Young's modulus of the substrate is approximately 220 GPa. After high oxidation, this value for the substrate with and without the TC layer is about 180 GPa and 160 GPa, respectively. The load–displacement curves for all specimens can be distinguished with three stages, namely, elastic, yielding, and plastic deformation stages as shown in Fig. 2. The figure clearly shows that the yield and maximum loads of TBC/substrate and bare substrate decrease rapidly because of the thermal exposure for 300 h, especially for the bare substrate. Moreover, the load bared by TBC/substrate during bending, especially after yielding, is higher than that of substrate. This difference increases after the high-temperature oxidation. These results indicate that the thermal barrier ceramic coating could enhance the resistance

713

to substrate deformation. The degradation of mechanical performance after high temperature oxidation is mainly caused by the microstructure change in the substrate but not the coating layer. Therefore, an abrupt decrease which appeared in the load–displacement curve of the pre-oxidized TBCs may be induced by this huge mismatch and stress release of the cracking. From the results of AE event counts recorded during the bending tests, as shown in Fig. 3, the numbers of AE signals produced from the bare substrate specimens are much less than those of the TBC/substrate. This suggests that the AE signals are mainly caused by the failure of the ceramic TC or the interface between coating and substrate. Three different regions with respect to the number of AE events for both the as-sprayed and the pre-oxidized TBC/substrate specimens were distinguished: (I) a very low number of AE events are pronounced at an initial loading time and ending at about 50 s and 150 s for as-sprayed and pre-oxidized TBC specimens. Based on the load–displacement curve, the specimen elastically deformed in this region; (II) AE activities maintain a relatively small and steady rate up to 590 s for the as-sprayed and 290 s for the pre-oxidized TBCs. In this region, the specimen undergoes elastic and plastic deformation. No visible interface debonding or coating spallation occurs, indicating failure in this region is mainly the surface vertical cracking; and (III) in this region, AE events rapidly increase due to a large release of the fracture energy associated with coating spallation. Thus the strong AE activities should be attributed to the

Time, s 600

600

0

240

480

720

960

TBC/substrate TBC/substrate Bare substrate

500

1200 420

a 350

a 400

360

280

300

210

Interface cracking

200 240

140

Surface cracking

Calculation for TBC/substrate Calculation for bare substrate Experiment for TBC/substrate Experiment for bare substrate

100

120

AE event

Load, N

Load, N

480

70

0 0.0

0.8

1.6

2.4

3.2

0 4.0

480

600 180

Displacement, mm 0 0.0

0.8

1.6

2.4

3.2

4.0

Time, s

Displacement, mm 240

0

200

b

200

120

240

360

Oxidized TBC/Substrate Oxidized TBC/Substrate Oxidized bare substrate

b 150

160

120

80

Calculation for TBC/substrate Calculation for bare substrate Experiment for TBC/substrate Experiment for bare substrate

40

0 0.0

120

0.8

1.2

1.6

2.0

90 Interface cracking Surface cracking

80

60

40

30

0 0.0 0.4

120

0.4

0.8

1.2

1.6

AE event

Load, N

Load, N

160

0 2.0

Displacement, mm

Displacement, mm Fig. 2. Load–displacement curves obtained from experiment and finite element simulation for the bare substrate and TBC/substrate: (a) as-sprayed and (b) pre-oxidized.

Fig. 3. Histories of load and AE event for the (a) as-sprayed and (b) pre-oxidized bare substrate and TBC/substrate, in which the bare substrate is nickel super-alloy, and the TBC/ substrate comprises a nickel super-alloy substrate as well as bond and ceramic coatings.

714

L. Yang et al. / Surface & Coatings Technology 232 (2013) 710–718

interface cracking. As shown in Fig. 2, the second region attributed to the surface cracking is much shorter for the pre-oxidized TBCs compared with that of the as-sprayed TBCs. In addition, the AE event count, load and duration time for the pre-oxidized specimens are much less than those of the as-sprayed specimen. Based on the idea that more cracking triggers more AE events, it can be concluded that high temperature oxidation decreases micro-failure sources but degrades the resistance to crack in TBC coating. To analyze fracture characteristics of TBCs under bending, the evolution of normal strain (along the x direction) in the TC layer and normal strain (along the y direction) at the coating/substrate interface for both the as-sprayed and the pre-oxidized TBCs were recorded by the DIC system. Fig. 4(a) shows an image with strain in the TC layer at 300 s for the as-sprayed TBCs, revealing that the tensile strain εTC xx for points a, b, c, d, e, f and g is up to 0.6%. At 800 s, several concentration regions of normal strain εBC yy are found, as shown in Fig. 4(b). The values at points h, i and j, are up to 0.5%, indicating that an interface crack has formed at each point for the as-sprayed TBCs. Fig. 4(c) schematically illustrates the failure process of the as-sprayed TBCs under bending. A surface crack in brittle ceramic coating forms and rapidly penetrates to the coating/substrate interface. Then, several surface cracks appear and a few of these kinks along the coating/substrate interface. With the increase of the bending loading, the normal strain εBC yy at the interface gradually ascends and results in interface cracking. The propagation and coalescence of interface cracks ultimately lead to the spallation of the coating. Similar to that which occurred in the TC

layer at 300 s as shown in Fig. 5(a), Fig. 5(b) illustrates that only one remarkable strain concentration region is found at the interface at 550 s for the pre-oxidized TBCs. The result implies that the failure of the pre-oxidized TBCs originates from the formation and propagation of one surface crack, followed by its kinking along the coating/ substrate interface to form one interface crack, and terminates with the spallation of the coating due to the propagation of this interface crack, as shown in Fig. 5(c). This result concurs with that of scanning electron microscopy observation, showing only one surface crack in the inserted figure. The characteristic of crack evolution can be analyzed by its AE feature as shown in Fig. 6. Several inflection points appear in the cumulative AE event curve for the as-received TBCs, indicating energy accumulation and abrupt release associated with the propagation of cracks and coalescence of each other. However, the cumulative AE event presents a smooth curve for pre-oxidized TBCs. 4.2. Surface fracture toughness and interface fracture toughness According to the AE characteristics as mentioned above, the fracture process of TBCs under bending can be distinguished with two stages, namely, surface cracking and interface cracking. Their initiation times or loads are important to understand the failure behavior and characterize the fracture toughness. Therefore, the curves of norBC mal strain εTC xx and εyy with the AE signals as well as the history of the load–displacement and AE events for the as-sprayed TBCs/substrate

c a

M

M Substrate

TC

BC

100 µm

a

-1.90

-1.00

-1.50

0.00

0.10

b

c d

0.20

e

0.30

g

f

0.75 [%]

0.40

b

100 µm

h

-0.60

-0.40

-0.30

-0.20

-0.10

0.00

i

0.10

j

0.20

0.50

[%]

Fig. 4. (a) Normal strain in the TC layer, (b) normal strain at the TC/BC interface, and (c) illustration of the failure process, of as-sprayed TBCs under three-point bending test.

L. Yang et al. / Surface & Coatings Technology 232 (2013) 710–718

715

Time, s

a

600

0

240

480

720

960

1200 6

8

5

400

Load, N

100 µm

4

300

3 Interface cracking

Surface cracking

200

2

Load AE event Normal strain ε xx

TC

B

100

1

Normal strain ε yy

6

4

Strain, %

500

Accumulative AE events, ×103

a

2

BC

A

-0.20

0.00

0.10

0.20

0.30

0.60

0.80

1.00 [%]

0 0.0

0.8

1.6

2.4

0 0 4.0

3.2

Displacement, mm Time, s

b

0

150

300

450

600 20

200

10

16

120

12

Load, N

100 µm

Surface cracking Interface cracking

80

8 Load

AE event

40 -1.60

0.00

0.50

1.00

1.50

2.00

2.50

3.00

C

[%] 0 0.0

c M

M

TC

D

4

Normal strain ε yy

6

4

2

BC

1.0

1.5

0 0 2.0

Displacement, mm Fig. 6. The cumulative AE events, the evolution of normal strain εTC xx at point e on the surface of coating shown in Fig. 4(a), the normal εBC yy at point h located at the interface of the ceramic coating and the bond coating shown in Fig. 4(b) for the (a) as-sprayed TBCs, and (b) pre-oxidized TBCs.

Substrate

TC

0.5

Normal strain ε xx

8

Strain, %

160

Accumulative AE events, ×102

b

TGO

BC

Fig. 5. (a) Normal strain in the TC layer, (b) normal strain at the TC/BC interface, and (c) illustration of the failure process, of the pre-oxidized TBCs under three-point bending test.

are shown in Fig. 6(a). Here, εTC xx is the strain evolution at point e shown in Fig. 4(a), εBC yy is the strain evolution at point h as shown in Fig. 4(b). Both AE events and the normal tensile strain history in the ceramic coating exhibit a break point at approximately 50 s, with εTC xx ≈ 0.35 %, a load of 300 N and deflection of 0.17 mm, corresponding to the initiation of surface cracking (point A) in Fig. 6(a). The AE events and the interface strain exhibit a break point at 590 s, with εBC yy ≈ 0.14 %, a load of 460 N, and deflection of 1.97 mm, corresponding to the initiative point of interface cracking (point B). Therefore, BC combining the evolution of εTC xx and εxy , the AE characteristic can be applied to analyze the critical initiative point of surface and interface cracks. Through similar analysis for the result of the pre-oxidized TBCs, as shown in Fig. 6(b), points C (~ 150 s) and D (~ 290 s) can represent the initiation time of surface and interface cracking for the pre-oxidized TBCs, respectively. The strain of coating at point C is 0.33%, load is 165 N, and the corresponding deflection is 0.52 mm. The relative strain of interface at point D is 0.19%, load is 145 N and deflection is 0.97 mm. Applying these critical thresholds and the values of the residual stress of −57 MPa and −200 MPa for the as-sprayed and the preoxidized TC layer, respectively, to Eq. (5), the surface fracture toughness

L. Yang et al. / Surface & Coatings Technology 232 (2013) 710–718

Table 1 The comparison of fracture toughnesses of TBCs with reference. Specimens

As-sprayed TBCs Pre-oxidation TBCs

Surface fracture toughness (KIC, MPa·m1/2)

Interface fracture toughness (KIIC, MPa·m1/2)

1.17, 0.95 ± 0.09 [10], 0.74 [28] 2.91, 3.06 [28]

2.13, 1.15 ± 0.07 [28] 2.27, 2.0 [25]

can be determined. The Young's modulus of TC layer and substrate that is obtained from the FEM simulation is 60 GPa and 220 GPa for as-sprayed TBCs, 80 GPa and 180 GPa for pre-oxidized TBCs,. The surface fracture toughness for the as-sprayed TBCs is 1.17 MPa·m1/2 or 22.82 J·m−2, which is approximated to 0.74 MPa·m1/2 measured by indentation [24], 0.95 ± 0.09 MPa·m1/2 from unilateral V beam test [10], and 25 J·m−2 from four-point bending test [9]. The surface fracture toughness for the pre-oxidation TBCs is 2.91 MPa·m1/2 or 105.85 J·m−2, which coincide with the result of 3.06 MPa·m1/2 [24] or 75 J·m−2 [9]. The critical load for interface cracking of TBCs before and after oxidation has been acquired exactly to be 460 N and 142 N. According to Eqs. (8) and (9), the interface fracture toughness for the as-sprayed and the pre-oxidized TBCs are 2.13 MPa·m1/2, and 2.27 MPa·m1/2, respectively. The critical energy release rate for the as-sprayed and the pre-oxidized TBCs are 51.82 J·m−2 and 23.52 J·m−2 respectively. As listed in Table 1, the results are approximated to the reported values [10,24]. 4.3. Correlation between cracking and AE parameters As mentioned above, AE signals originating from substrate can be neglected. Thus, the AE energy recorded before interface cracking is merely related to surface cracking and the conversion coefficient αs in Eq. (7). Four different loads (358 N, 428 N, 449 N, 459 N) in the range of the surface cracking process (300 N ~ 460 N) are chosen to analyze the relationship between cracking and AE parameters. After being bent at each load, the number of surface cracks is carefully counted under an optical microscope. Then, energy released from the surface cracks can be calculated by Eq. (6) to analyze its correlation with the corresponding AE signals. As shown in Fig. 7, a linear relationship exists between fracture energy released by the surface cracks in the TC layer and the corresponding AE signals, which has a slope αs of about 1.714 × 108. The results of the crack number N, fracture energy EFS, AE energy ESAE, and proportional coefficient αs are listed in Table 2. The number of the cracks increases with the loading. It is worth noting that the relationship between fracture and AE energies was not considered owing to the fact that there is only one surface crack for the pre-oxidized TBCs.

Table 2 Summary of the AE energy (ESAE) and the total energy (EFS) generated by surface cracks for as-sprayed specimens. As-sprayed specimens P (N) N EFS (10−4 J) ESAE (10−12 J) αs (108)

1#

2#

3#

4#

358 1 0.7272 0.3806 1.9107

428 3 2.1816 1.7014 1.2825

449 5 3.6360 2.2611 1.6080

459 6 4.3632 2.2779 1.9154

The failure process of TBCs originates from surface cracking, and then proceeds by delamination under the bending condition. Therefore, AE signals generated after the critical load of the interface crack can be regarded as the response to delamination. Similarly, four different loads in their range of interface cracking are chosen to investigate the relationship of energies from fracture and AE signals for both the assprayed and pre-oxidized TBCs. The length of interface cracks can be determined by SEM and the energy released from the interface cracks EFI can be calculated by using Eq. (8). A linear relationship of energy released from interface crack and AE signals was also found, for the as-sprayed and pre-oxidized TBCs, respectively, as shown in Fig. 8(a) and (b). Compared with Fig. 7, the AE energy generated by the interface cracking is larger than that of the surface cracking. The results of crack length a, fracture energy EFI, AE energy EIAE and proportional coefficient αI are listed in Table 3. For the as-sprayed TBCs, the slope of the linear line αI is 1.35 × 108 for the interface crack, whereas the value is approximately 2.21 × 108 for that of pre-oxidized TBCs. Therefore, it can be concluded that there is a linear relationship between the energy 2.5

a 2.0

EFI, 10-3 J

716

Experimental data Fitting

1.5

1.0

0.5

0.0 0.0

0.3

0.6

0.9 I

1.2

1.5

-11

EAE, 10 J 1.5

5 1.2

Experimental data Fitting

-3

EFI, 10 J

EFS, 10-4J

4

3

2

0.9

0.6

0.3

1

0 0.0

b

Experimental Data Fitting

0.6

1.2

1.8 S

2.4

3.0

-12

EAE , 10 J Fig. 7. Relationship between energy released from the surface cracking in the as-sprayed TBCs and the corresponding AE signals.

0.0 0.0

1.2

2.4

3.6 I

4.8

6.0

-12

EAE, 10 J Fig. 8. Relationship between energy released from the interface cracking and the corresponding AE signals for the: (a) as-sprayed TBCs and (b) pre-oxidized TBCs.

L. Yang et al. / Surface & Coatings Technology 232 (2013) 710–718

717

Table 3 Summary of the AE energy (EIAE) and the total energy (EFI) generated by interface cracks for as-sprayed specimens. Parameters

As-sprayed specimens

P (N) a (mm) EFI (10−4 J) EIAE (10−12 J) αI (108)

470 0.645 6.43 3.70 1.74

Pre-oxidation specimens 475 0.873 8.99 7.42 1.21

480 1.372 14.58 11.21 1.30

released from fracture and the corresponding AE signals. The slope is related to AE acquisition system and material properties, such as the microstructure change induced by high temperature oxidation. 4.4. Influence of high temperature oxidation on fracture behavior of TBCs SEM images of cross section of the as-sprayed and oxidized specimens are shown in Fig. 9(a) and (b), respectively. The TC layer has a typical APS microstructure, including sprayed particles, pores, delamination and oxide. These microdefects in the TC layer that are generally regarded as sources of surface cracks decrease after high temperature oxidation due to sintering of ceramic. Therefore, different to the as-sprayed TBCs, only one surface crack occurred for the pre-oxidized TBCs, resulting in only one interface crack formation and propagation with the smooth increase of the AE signals. Through the calculation discussed in Section 4.2, the surface fracture toughness is 1.17 MPa·m1/2 and 22.82 J·m−2 for the as-sprayed TBCs, which increases to 2.91 MPa·m1/2 and 105.85 J·m−2 after high temperature oxidation. Therefore, the microstructure change in the TC layer after high temperature oxidation improves the resistance to surface crack, characterized by less surface cracks and corresponding AE signals. As shown in Fig. 9, a rough interface was prepared to improve the mechanical bonding of TC to the BC layer. After high temperature

a

485 1.784 19.57 13.31 1.47

144 0.952 4.23 1.62 2.61

146 1.258 5.54 2.69 2.06

148 2.072 8.82 4.09 2.16

150 2.937 11.99 5.05 2.37

oxidation, the TC/BC interface, BC/substrate interface and the BC layer are distributed with dark oxides, and the thicknesses of TGO are about 6 and 2 μm in TC/BC interface and BC/substrate interface respectively, as shown in Fig. 9(c). This oxidation characteristic has been found experimentally in our previous work [28]. In this case, the thickness of each TGO layer and the number of layers in the BC layer increase with the exposure time, but continuous TGO layers only occur after 200 h oxidation. To clearly show the difference between as-sprayed and preoxidized, the pre-oxidation treatment at 1000 °C for 300 h was investigated in this paper. Commonly, the adhesive strength of BC to TC is less than that to substrate for as-sprayed TBCs. Therefore, the spallation of coating occurs at the TC/BC interface for as-sprayed TBCs, as shown in Fig. 4. After high oxidation, however, the fracture behavior of TBCs is affected by the characteristic of TGO. Commonly, a thick TGO layer over 10 μm formed at the ceramic/bond coating interface [2,6,15,29]. In this case, the spallation or delamination appears at the interface between TC layer and TGO or TGO and BC layer under both mechanical testing and thermal cycling [2,6,29]. When TGO formed in the whole BC layer, especially at the BC/substrate interface, the adhesive strength of the BC/substrate interface sharply descended due to the fact that adhesion between ceramic TGO and metal substrate is lower than that between ceramic TGO and ceramic ZrO2 layer. In this case, the spallation or delamination appears at the BC/substrate interface [20], as shown in Fig. 5. To verify the phenomena, the traditional tensile test is

b

Epoxy

Microcrack Pore

Ceramic coating

S EI

c

30kV

WD18mm

S S 30

×300

50µm

d Epoxy Ceramic coating TGO

Bond coating SEI

30kV

WD18mm

SS30

×300

50µm

Fig. 9. The micromorphology of cross sections: (a) as-sprayed and (c) pre-oxidized TBCs; (b) and (d) spalled coating after tensile for as-sprayed and pre-oxidized TBCs, respectively.

718

L. Yang et al. / Surface & Coatings Technology 232 (2013) 710–718

conducted to obtain spallation interface location and corresponding adhesive strength for as-sprayed and pre-oxidized TBCs. Fig. 9(b) shows that the interface debonding occurs in TC/BC interface for as-sprayed TBCs, however, it occurs in BC/substrate interface due to a thickness of 2 μm of TGO for pre-oxidation TBCs, as shown in Fig. 9(d). The adhesive strength of TC/BC interface for as-sprayed TBCs and BC/substrate interface for pre-oxidized TBCs is 47.2 and 16.8 MPa, respectively. Obviously the adhesive strength of the interface between the bond coat and the substrate is more than that (47.2 MPa) between the top and bond coat for as-sprayed TBCs because the spallation occurs at the TC/BC interface. Therefore, very thin TGO sharply decreases the adhesive strength of BC/substrate interface with the value of 16.8 MPa, far lower than that of TC/BC interface. Therefore, the fact that the interface crack appears at the BC/substrate interface during the bending test for pre-oxidation TBC specimen is reasonable. The experimental results also demonstrate that TGO formed in BC/substrate interface is more dangerous for TBCs. For as-sprayed TBCs, delamination occurs at the TC/BC interface and thus the fracture toughness of 2.13 MPa·m1/2 and 51.82 J·m−2 determined according to Eqs. (8) and (9) is to evaluate the resistance of TC/BC interface crack. However, crack appears at the BC/substrate interface for the pre-oxidized TBCs with the fracture toughness of BC/ substrate at 2.27 MPa·m1/2 and 23.52 J·m−2. Due to the difference in the dimensionless parameters of A, p, I, ∑, I0, Δ, η and γ between the BC/TC and BC/substrate interfaces, the fracture toughness expressed by KIIC and GIIC between the as-sprayed and pre-oxidation TBCs is not strictly comparable. The decrease of critical load (from 460 N to 145 N) for the initiation of interface cracking after high temperature oxidation, as shown in Fig. 2, is mainly caused by the deterioration of substrate (general term of bond coating and nickel-based substrate) performance such as Young's modulus. These results reveal that thermal exposure has an important influence on failure mechanism of TBCs prepared by the air plasma spraying method. As discussed in Section 4.3, a linear relationship between the energy released from cracking and corresponding AE signals for both as-sprayed and pre-oxidized TBCs, with the slope being dependent on the failure type of TBCs. The slope increases after high temperature oxidation of TBCs, indicating that the oxidation decreases the energy scattering caused by the transmission through TBCs. 5. Conclusions The fracture characteristics of TBCs under bending have been investigated using AE monitoring. The main conclusions can be summarized as follows: (1) Details on fracture behavior of TBCs can be analyzed through their AE features. (2) The critical crack-driven load can be accurately determined by the combination of AE signals and strain evolution. These critical values can be used to characterize the fracture toughness of TBCs. The results show that for the as-sprayed and pre-oxidized

TBCs, the surface fracture toughnesses are 1.17 MPa·m1/2 and 2.91 MPa·m1/2, respectively, and the interface fracture toughnesses are 2.13 MPa·m1/2 and 2.27 MPa·m1/2, respectively. (3) A linear relationship exists between energy released from cracking and the corresponding AE signal for both the as-sprayed and pre-oxidized TBCs, whose slope depends on the failure type and microstructure evolution of TBCs. (4) High temperature oxidation increases the resistance of TBCs to surface cracking. However, when a very thin TGO layer formed at the BC/substrate interface, the adhesive strength of the BC/ substrate interface sharply descended and is much smaller than that of TC/BC interface. In this case, the spallation or delamination appears at the BC/substrate interface. Acknowledgments This work has been supported by the National Natural Science Foundation of China (Nos. 11002122, 51172192, 11272275, 11002121 and 10828205), the Natural Science Foundation of Hunan Province (No. 11JJ4003) and the Key Project of Scientific Research Conditions in Hunan Province (No. 2012TT2040). The specimens were provided by the AVIC Shenyang Liming Aero-Engine (GROUP) Corporation Ltd. References [1] A.G. Evans, D.R. Mumm, J.W. Hutchinson, G.H. Merier, F.S. Petit, Prog. Mater. Sci. 46 (2001) 505–553. [2] N.P. Padture, M. Gell, E.H. Jordan, Science 296 (2002) 280–284. [3] R.A. Miller, Surf. Coat. Technol. 30 (1987) 1–11. [4] G.C. Chang, W. Phucharoen, R.A. Miller, Surf. Coat. Technol. 30 (1987) 13–28. [5] D.S. Balint, J.W. Hutchinson, J. Mech. Phys. Solids 53 (2005) 949–973. [6] S.R. Choi, J.W. Hutchinson, A.G. Evans, Mech. Mater. 31 (1999) 431–447. [7] M.R. Begley, D.R. Mumm, A.G. Evans, J.W. Hutchinson, Acta Mater. 48 (2000) 3211–3220. [8] S.S. Kim, Y.F. Liu, Y. Kagawa, Acta Mater. 55 (2007) 3771–3781. [9] Y. Yamazaki, A. Schmidt, A. Scholz, Surf. Coat. Technol. 201 (2006) 744–754. [10] S.R. Choi, D. Zhu, R.A. Miller, Ceram. Eng. Sci. Proc. 19 (1998) 293–301. [11] Y.C. Zhou, T. Hashida, C.Y. Jian, J. Eng. Mater. Technol. 125 (2003) 176–183. [12] E.P. Busso, L. Wright, H.E. Evans, L.N. McCartney, S.R.J. Saunders, Acta Mater. 55 (2007) 1491–1503. [13] X. Chen, J.W. Hutchinson, M.Y. He, A.G. Evans, Acta Mater. 51 (2003) 2017–2030. [14] R. Vaßen, G. Kerkhoff, D. Stöver, Mater. Sci. Eng., A 303 (2001) 100–109. [15] E.P. Busso, H.E. Evans, Z.Q. Qian, M.P. Taylor, Acta Mater. 58 (2010) 1242–1251. [16] D. Renusch, M. Schütze, Surf. Coat. Technol. 202 (2007) 740–744. [17] X.Q. Ma, M. Takemoto, Mater. Sci. Eng., A 38 (2001) 101–110. [18] L. Fu, K.A. Khor, H.W. Ng, T.N. Teo, Surf. Coat. Technol. 130 (2000) 233–239. [19] X.Q. Ma, S. Cho, M. Takemoto, Surf. Coat. Technol. 139 (2001) 55–62. [20] L. Yang, Y.C. Zhou, W.G. Mao, Q.X. Liu, Surf. Interface Anal. 39 (2007) 761–769. [21] L. Yang, Y.C. Zhou, W.G. Mao, C. Lu, Appl. Phys. Lett. 93 (2008) 231906. [22] W.B. Yao, C.Y. Dai, W.G. Mao, C. Lu, L. Yang, Y.C. Zhou, Surf. Coat. Technol. 206 (2012) 3803–3807. [23] L. Yang, Y.C. Zhou, C. Lu, Acta Mater. 59 (2011) 6519–6529. [24] Q. Chen, W.G. Mao, Y.C. Zhou, C. Lu, Appl. Surf. Sci. 256 (2010) 7311–7315. [25] G. Thurn, G.A. Schneider, H.-A. Bahr, F. Aldinger, Surf. Coat. Technol. 123 (2000) 147–158. [26] Z. Suo, J.W. Hutchinson, Int. J. Fract. 43 (1990) 1–18. [27] S.Q. Guo, Y. Kagawa, Scripta Mater. 50 (2004) 1401–1406. [28] L. Yang, Y.C. Zhou, W.G. Mao, Surf. Rev. Lett. 15 (2007) 935–943. [29] J. Rösler, M. Bäker, K. Aufzug, Acta Mater. 52 (2004) 4809–4817.