Surface and Coatings Technology 157 (2002) 118–127

Fracture characteristics of thermal barrier coatings after tensile and

bending tests
Y.C. Zhoua,b,*, T. Tonomoria, A. Yoshidaa, L. Liua, G. Bignallc, T. Hashidaa
a
Fracture Research Institute, Tohoku University, Aoba 01, Aramaki, Aoba-ku, Sendai, 980-8579, Japan
b
Institute of Fundamental Mechanics and Material Engineering, Xiangtan University, Xiangtan, Hunan 411105, PR China
c
Environmental Geochemistry Laboratory, Department of Geoscience and Technology, Tohoku University, Aoba 01, Aramaki, Aoba-ku, Sendai,
980-8579, Japan

Received 8 August 2001; accepted in revised form 29 March 2002

Abstract

Uniaxial tension and four-point bending tests were conducted on two-layer (non-functionally graded material) plasma-sprayed
rectangular specimens, boasting a new, functionally graded material (FGM) layer. The top coat layer was composed of stabilized
yttria-zirconia (YSZ), and the bond coat layer was NiCrAlY. The FGM was composed of five layers: 100% YSZ; 80% YSZq
20% NiCrAlY; 60% YSZq40% NiCrAlY; 40% YSZq60% NiCrAlY; and 20% YSZq80% NiCrAlY. Fracture in the thermal
barrier coating (TBC) system was examined by scanning electron microscopy (SEM), which showed that vertical multiple
cracking first occurred in the top coat, followed by propagation of interface cracking between the top and bond coats. Spallation
of non-FGM coatings occurred in tensileybending experiments, but not for the FGM-coated specimens, with only localized
delamination observed along the interface between FGM layers and the substrate. For specimens prepared with a gun–substrate
distance of approximately 120 mm and gun operating power of approximately 32.5 kW, the interface fracture toughness, evaluated
by a shear lag model for the uniaxial tensile test, was 0.94 and 0.67 MPa m1y2 for type-A and -B coatings, respectively. Here,
types A and B correspond to an average roughness of the bond coat surface of 12.8 mm and 6.8 mm, respectively. The interface
fracture toughness evaluated by the Suo–Hutchinson model was 1.05–1.27 and 1.0–1.17 MPa m1y2 for type-A and -B coatings,
respectively. The results are very close to data obtained by the blister test method. For specimens prepared with a gun–substrate
distance of 125–130 mm and gun operating power of approximately 38.5 kW, the interface fracture toughness, evaluated by the
Suo–Hutchinson model for four-point bending, was 4.26–7.21 MPa m1y2. 䊚 2002 Elsevier Science B.V. All rights reserved.

Keywords: Zirconia; Plasma spraying; Fracture characteristics; Scanning electron microscopy; Thermal barrier coating

1. Introduction w1x. The program will require development of: (1) high-
performance, high-reliability turbine vanes and blades;
There is strong scientific evidence that the present (2) efficient hot-parts cooling technology; and (3) a
level of atmospheric carbon dioxide emission should be high-temperature and low-NOx combustor.
reduced, since it is a primary cause of global warming. Materials of high performance and high reliability
In the case of fossil fuel-fired power generation, higher include new (high) heat-resistant alloys and thermal-
thermal efficiency is most effective in reducing carbon barrier ceramic coatings (TBCs). The aim of developing
dioxide emission, and the ‘Development of the TBC-related technology is to protect a heat-resistant
Advanced Gas Turbine’ will lead to electric power alloy, to ensure its effectiveness at 1500 8C, and to
systems with few detrimental effects on the environment realize 50% thermal efficiency w2–9x. Defects such as
delamination, spallation, or cracking due to thermal
*Corresponding author. Present address: Institute of Fundamental stress in TBCs are critical events that disrupt continuous
Mechanics and Material Engineering, Xiangtan University, Xiangtan,
Hunan 411105, PR China. Tel.: q86-732-8293586; fax: q86-732-
operation. In particular, coating failure by spalling,
8292468. accompanied by delamination, has a crucial influence
E-mail address: yichunzhou@hotmail.com (Y.C. Zhou). on the function and life expectancy of a thermal barrier.

0257-8972/02/$ - see front matter 䊚 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 2 5 7 - 8 9 7 2 Ž 0 2 . 0 0 1 5 4 - 8
 Y.C. Zhou et al. / Surface and Coatings Technology 157 (2002) 118–127 119

Fig. 1. Dimensions of TBC specimens in tension experiments, where t is the substrate thickness.

Many workers have considered the life prediction and 60% NiCrAlY; and (e) 20% YSZq80% NiCrAlY. The
failure mechanism of TBC systems w2–9x and found thickness of the FGM coating was approximately 1.0
that interfaces, due to the presence of defects, tend to mm, with the coating distribution determined by a
be planes of weakness. Due to mismatch of the mechan- computer-controlled plasma gun. Coating specimens
ical and thermal properties, interfacial defects and rough- were marked as N and T, corresponding to a gun–
ness commonly experience severe stresses during long substrate distance of approximately 120 and 125–130
service. In recent years, several investigations related to mm and gun operating power of approximately 32.5 and
bi-material fracture mechanics have been undertaken 38.5 kW, respectively.
that specifically address interface cracking related to Many researchers w11x have shown that the roughness
mechanical and thermal loading, as reviewed by Hutch- of the bond coat has important effects on interface
inson and Suo w10x. characteristics, and consequently on the operating life
In this paper, we describe a method for determining of the TBC ceramic coating. For this reason, bond coat
the fracture characteristics of a ceramic top coatybond roughness was measured with a Surfcorder SE-2300,
coat interface. Uniaxial tension and four-point bending and is defined here by the following formula:
tests were conducted on plasma-sprayed, rectangular L
1
specimens, comprising a top coat layer of yttria-stabi-
lized zirconia (YSZ), a bond coat layer of NiCrAlY and
Ras
L |Z0
fŽx.Zdx (1)

five layers of functionally gradient material (FGM) as where L is the length of the measured region, and f (x)
the TBC system. The fracture process in the thermal is the deviation about the center line.
barrier coating system was examined by scanning elec-
tron microscopy (SEM). The interface fracture tough- 2.2. Tensile experiment
ness for the non-functionally gradient material
(non-FGM) TBC system is based on experimental data Tensile experiments were carried out on specimens
for bending and tensile experiments. comprised of a two-layer TBC system on a stainless
steel (SUS304) substrate. The dimensions of the tensile
2. Experimental procedure specimens are shown in Fig. 1. The thickness of the top
and bond coats and the average roughness of the bond
2.1. Specimen preparation coat surfaces are listed in Table 1.
Specimens were loaded along their longitudinal axis,
In our experiments, two types of material were used as shown in Fig. 1, with displacement measured using
as substrates: stainless steel (SUS304) and carbon steel a linear variable differential transducer (LVDT). The
(S45C). Furthermore, two types of ceramic TBC were span of displacement measured was 50 mm. The loading
investigated: a general two-layer coat system (i.e. non- rate of 0.12 mmØminy1 was governed by the displacing
functionally gradient material, non-FGM), and one com- rate.
prised of a FGM coating.
For both systems, NiCrAlY was deposited onto the 2.3. Four-point bending experiment
substrate by plasma spraying as a bond coat, with a
thickness of approximately 100 mm. For the non-FGM Four-point bending experiments were conducted with
system, a 440-mm-thick YSZ (zirconia stabilized with specimens on stainless steel (SUS304) and carbon steel
8 wt.% Y2O3) top coat was deposited onto the bond (S45C) substrates, with a two-layer coat system or,
coat by plasma spraying. The FGM system consisted of alternatively, an FGM coating. The dimensions of spec-
five layers: (a) 100% YSZ; (b) 80% YSZq20% Ni- imens tested using four-point bending are shown in Fig.
CrAlY; (c) 60% YSZq40% NiCrAlY; (d) 40% YSZq 2. Table 2 lists the thickness values of the substrate,
120 Y.C. Zhou et al. / Surface and Coatings Technology 157 (2002) 118–127

Table 1
Dimension and thickness of TBC specimens in tension experiments

Substrate Sample Substrate Bond coat Topcoat Average

symbol thickness thickness thickness roughness
(mm) (mm) (mm) of bond coat
surface (mm)
Stainless steel (SUS 304) A 1.5 100 440 12.8
B 1.5 100 440 6.8
A 3.0 100 440 12.8
B 3.0 100 440 6.8

Fig. 2. Dimensions of TBC specimens in four-point bending experi-

ments, where t is the substrate thickness.

bond coat and top coat. Specimens were loaded into an

Fig. 3. Four-point bending apparatus, where L9s90 mm is the center-
Instron machine with a four-point fixture, as shown in to-center distance between the lower loading point, and Ls30 mm is
Fig. 3. In the Instron machine, the upper loading part the center-to-center distance between the upper loading points, d is
was fixed (i.e. A). Loading was achieved by moving the relative displacement from the midway point of the specimen to
the lower part of the specimen (i.e. B). The strain (´) part B, and was measured by laser beam. D is the displacement of
and curvature radius (r) of the deformed specimen were part B, and was measured by a linear variable differential transducer
(LVDT).
determined by the following expressions:
h1 miny1. The substrate thickness (t) is much larger than
´s (2)
r the bond coat layer thickness (hb), and it was assumed
ŽDyd.2qŽLy2.
2 that the neutral axis (located midway between the
rs qh1 (3) substrateybond coat interface and the substrate rear side)
2ŽDyd. was parallel to the bottom face of the specimen.
where L is the span of upper loading points and h1 is
the half-substrate thickness. In Eq. (3), d is the relative 2.4. Post-testing investigations
displacement from the midway position in the specimen
to part B, and was measured by laser beam (Fig. 3). D To characterize the development of the microstructure,
is the displacement of part B, and was measured by cross-sections of tension and bending-tested specimens
LVDT. The Instron machine had a load point displace- were observed by FE-SEM S-4700 scanning electron
ment rate of 0.18 mm miny1, i.e. dDydts0.18 mm microscopy (SEM). Cross-sections were prepared by

Table 2
Dimension and thickness of TBC specimens in four-point bending experiments

Coating Substrate type Substrate Bond coat Topcoat

thickness thickness thickness
(mm) (mm) (mm)
Non-FGM Stainless steel (SUS304) 1.5 100 440
3.0 100 440
5.0 100 440
Carbon steel 3.0 100 440

FGM Stainless steel (SUS304) 3.0 100 1000

 Y.C. Zhou et al. / Surface and Coatings Technology 157 (2002) 118–127 121

in eventual spallation of the top coat ceramic layer (Fig.

6c).
In some cases, the vertical crack propagated through
the bond coat layer and delamination took place at the
bond coatysubstrate interface, as shown in Fig. 6a. The
initial cracks were observed on the surface of top coat
layer when the strain of substrate was about 3.0, 3.12,
2.49 and 2.0% for (N, A, ts1.5 mm), (N, A, ts3.0
mm), (N, B, ts1.5 mm), (N, B, ts3.0 mm) specimens,
respectively. Here, type N corresponds to a gun–sub-
strate distance of approximately 120 mm and gun
operating power of approximately 32.5 kW; types A and
B correspond to an average roughness of the bond coat
surface of 12.8 and 6.8 mm, respectively; and t denotes
the thickness of the substrate. It was noted that the
Fig. 4. Stress–strain curve for non-FGM TBC specimens from our
initial cracks appeared when the substrates were loaded
tension experiment with a (1.5 mm thick) substrate of stainless steel in the plastic region. It is also evident that the strain
(SUS 304). N corresponds to a gun–substrate distance of approxi- required to induce initial cracking for high bond-coat
mately 120 mm and gun operating power of approximately 32.5 kW; roughness was higher than that for low roughness. In
t denotes the substrate thickness. The bond coat roughness RA for type the experiment, spallation of the top coat ceramic layer
A coating is 12.8 mm.
occurred immediately after reaching the largest strain.
The strain required to induce delamination at the top
wet diamond sawing at a low speed to minimize speci- coatybond coat interface decreased with increasing sub-
men damage. One side of the specimen was polished in strate thickness (Fig. 5). In Fig. 5, it is clear that crack
the upper loading span, L, using emery polishing paper density decreases with decreasing bond coat roughness.
(1500-, 900- and finally 600-grit). Subsequently, sequen- In order to obtain quantitative knowledge about the
tial polishing was conducted using a soft disk emery quality of ceramic TBC systems, a simple shear lag
suspension, with the suspension size being 15, 9, 6, 3 model was used to evaluate interface strength. Hu and
and finally 1 mm. This procedure was conducted at Evans w12x showed that when the substrate is ductile,
room temperature. the cracked coating either remains attached to the
substrate or decoheres along the interface, depending
3. Results and discussion upon coating adhesion and substrate yield strength. A
critical non-dimensional parameter (Vc) and a relative
3.1. Tensile test critical cracking stress (sc yY) were defined to evaluate

Fig. 4 shows the stress–strain curve measured for

specimen with coating type A. Fig. 5 shows the crack
density, which is defined as the ratio of the number of
cracks on the top coat surface to specimen length, vs.
strain, for type A and B coatings having high and low
roughness of the bond coat surface, respectively, as
described in Table 1.
The fracture process in the tensile test can be sum-
marized as follows: cracks normal to the loaded axis
appeared on the surface of the top coat layer, as shown
in Fig. 6a. These cracks are referred to as ‘surface’ or
‘vertical’ cracks. This phenomenon is the same as that
observed by Hu and Evans w12x and Yanaka et al. w13x
for brittle ceramic coating. However, when the coating
is ductile, surface cracks occur at a direction of 458
angle to the load axis w14x. When the surface crack tip
is close to the top coatybond coat interface, the crack Fig. 5. Crack density vs. strain for non-FGM TBC specimens from
kinks and propagates along this interface (Fig. 6b). our tensile test with a substrate of stainless steel (SUS 304). N cor-
responds to a gun–substrate distance of approximately 120 mm and
These cracks are referred to as ‘delamination’ cracks. gun operating power of approximately 32.5 kW; t denotes the sub-
Discontinuous delamination cracks that formed at the strate thickness. The bond coat roughness RA for type A coating is
top coatybond coat interface tend to link up, resulting 12.8 mm and for type B coating is 6.8 mm.
122 Y.C. Zhou et al. / Surface and Coatings Technology 157 (2002) 118–127

Fig. 6. SEM cross-section for specimens with a substrate thickness of 1.5 mm after tension testing, where the bond coat layer roughness was
RAs12.8 mm. (a) Vertical cracking; (b) vertical cracking and interface cracking; and (c) link-up of discontinuous delamination cracks at the
topybond coat interface, resulting in spallation of the topcoat.

cracking and decohesion of the thin coating (h<t) on Lct

a ductile substrate, where sc and Y are the fracture scs (4)
h
strength of the coating and the yield strength of the
substrate, respectively. 1 t
PLs s (5)
Generally, when the relative critical cracking stress Lc hsc
sc yY is greater than 0.2, the shear lag model may be
used to estimate interface strength w12x. In our investi- where Lc is the crack spacing and h is the top coat
gation, the fracture strength of ceramic coat material thickness (Fig. 7), PL is the crack density and t is the
(sc) was 100 MPa w15x, and the substrate NiCrAlY interface shear strength. The crack density can be read
yield strength (Y) was 426 MPa w16x. Consequently, the from Fig. 5 and is 0.16–0.18 and 0.1 mmy1 for type A
relative critical cracking stress (sc yY) is 0.2347, which and B coatings, respectively. Using Eq. (5), this results
is greater than 0.2, so the shear lag model is valid. The in ts7.0 (type A coating) and 3.6 MPa (type B
governing equations w12x are: coating). The interface fracture toughness can be cal-
 Y.C. Zhou et al. / Surface and Coatings Technology 157 (2002) 118–127 123

Fig. 7. Schematic representation of shear lag behavior used to model

TBC cracking at the bond coatytop coat interface.

culated by:
KcstypLc (6)
Fig. 8. Stress intensity factor vs. strain for non-FGM TBC specimens
1y2
This results in Kcs0.94 and 0.67 MPa m for type A in our tensile test with substrate of stainless steel (SUS 304). N cor-
and B coatings, respectively, which are very close to the responds to a gun–substrate distance of approximately 120 mm and
data obtained by blister test method w17,18x. gun operating power of approximately 32.5 kW; t denotes the sub-
strate thickness. The bond coat roughness RA for type A coating is
The interface fracture toughness is discussed using
12.8 mm and for type B coating is 6.8 mm.
the Suo–Hutchinson analysis on interface cracking
between two elastic layers w19x. The energy release rate
PsC1Q Ms0 (10)
with the Suo–Hutchinson formula wEq. (A1)x w19x is
given in Appendix A: Therefore, the stress intensity factor for tensile experi-
ment can be obtained from Eq. (7), Eq. (8) and Eq.
c1 w P2 M2 PM z
(10) using data in Fig. 4. The results are shown in Fig.
Gs x q 3 q2 sing| (7)
16 y Ah Ih yAIh2 ~ 8.
The initial rise in the K–resistance curve is due to
Elastic parameters, such as Young’s modulus and Pois- the growth of the so-called ‘fracture process zone’, as
son’s ratio for the coating, bond coat and SUS304 steel discussed in w18x. The stress intensity factor, shown in
are listed in Table 3 w18x. The elastic parameters for the Fig. 8 for complete spallation, is regarded as the fracture
NiCrAlY bond coat and stainless steel SUS304 are toughness of the interface. It is evident that the interface
similar. Therefore, the combination of bond coat and fracture toughness is Kcs1.05–1.27 and 1.0–1.17 MPa
SUS304 stainless steel is regarded as a substrate. The m1y2 for type A and B coatings, respectively, which are
complex stress intensity factor K is written as: close to values estimated by the shear lag model dis-
4coshp´ cussed above.
ZKZs yG (8) It is clear that vertical cracks correspond to a higher
yc1qc2 interface strength (from Fig. 5 and the results of inter-
Here, the stress intensity factor of the interface crack face fracture toughness). The phenomenon was also
associated with tensile experiments for thermal barrier indicated by Yanaka et al. w13x. For a given load, longer
ceramic coating is equivalent to that induced by the segments experience the largest tensile stresses at their
following load and moment, as given in Appendix A: centers, and break into (more or less) ‘half lengths’,
while shorter segments are unbroken, as discussed by
P1s0 M1s0 P3syQ M3s0 (9) Yanaka et al. w13x. The stress experienced by the long
The stress Q associated with strain is known from Fig. segments, with ls4Ly3 ¯ , is just below the critical fracture
4. Substitution of Eq. (9) into Eq. (A3) gives: strength (where L̄ is the mean crack spacing w13x). In

Table 3
Material parameters

Mechanical parameters PSZ NiCrAlY SUS304

Type A Type B
2
Ey(1yn ) (GPa) 18.6 (16.4–20.3) 9.7 (9.5–11.4) 211 209
E (GPa) 18.5 9.7 192 190
n 0.08 0.08 0.3 0.3
124 Y.C. Zhou et al. / Surface and Coatings Technology 157 (2002) 118–127

chemical bonding. An optimum roughness should be

found, which minimizes interfacial stress, whilst facili-
tating mechanical bonding.

3.2. Four-point bending test

The curve of load vs. strain is shown in Fig. 9 for

the four-point bending tests of the non-FGM system,
with stainless steel substrate of different thickness (as
listed in Table 2). Fig. 10 shows the change in crack
density with strain for non-FGM and FGM coating
systems, with a stainless steel or carbon steel substrate.
Crack density, PL, is defined by the ratio of multiple
crack numbers to the distance L (between upper loading
points). In our experiments, spallation of the top coat
ceramic layer occurred immediately upon reaching the
largest strain level. It was found that, when the substrate
Fig. 9. Load–strain curve for the non-FGM coating system with stain- had been loaded in the plastic region, (with strain of
less steel substrate during the four-point bending test; t denotes the
0.2%), the cracks became prominent and propagated
substrate thickness. T denotes the specimen conditions, i.e. gun–sub-
strate distance of 125–130 mm and gun operating power of approx- through the ceramic layer (Fig. 10).
imately 38.5 kW. Interface fracture toughness is assessed by Suo–
Hutchinson analysis of interface cracking between two
long segments, if the interface strength is high, then the elastic layers w19x. The stress intensity factor of interface
interface crack will not initiate, and a vertical crack will cracking, associated with four-point bending of the
be produced. On the other hand, if the interface strength thermal barrier ceramic coating, is exactly equivalent to
is low, then the vertical crack will not be produced and that induced by load and moment combinations given
interface delamination will result. In this case, more by Suo and Hutchinson w19x:
vertical cracks correspond to greater interface strength. P1s0 M1s0 P3s0 M3syP0(L9yL)y4b (11)
It is evident that the interface strength for type A
where b is the width of the specimen. The total load
coating with a high-roughness bond coat is greater than
P0 associated with the strain is known from Fig. 9.
for type B coating with a low-roughness bond coat. We
Substitution of Eq. (11) into Eq. (A3) gives:
conclude that the roughness of the bond coat surface
has an important effect on the interface strength of the P0(L9yL) P0(L9yL)
PsC2 MsC3 (12)
TBC system. As discussed, crack density increases with 4bh 4b
increasing bond coat roughness. This means that the
interface strength increases with increasing bond coat
roughness. A certain degree of roughness is necessary
to promote a mechanical bond between the top coat and
the bond coat w20x, because chemical bonding between
the two materials is minimal. In our experiment, cracks
originated in the top coat surface. This means that a
certain degree of roughness of the bond coat will induce
micro-cracking in the top coat surface. However, it is
known that the effect of bond coat roughness on the
stress field in the top coat is limited to the interface
region w21,22x. High tensile stresses occur normal to the
interface between the bond coat and top coat, in the
vicinity of the peak of the sine wave (the tip of the
asperity). Such tensile stresses are responsible for initi-
ating cracks in the ceramic layer, in a critical zone in
the ceramic, several mm away from the sine peak. This
is the conclusion drawn by Petrus and Ferguson w20x: a
rough interface provides stress concentration sites at the
interface between the materials. A smooth interface
Fig. 10. Plot of crack density vs. strain for four-point bending test of
would have no stress concentration sites, but bonding non-FGM and FGM coating systems, with stainless steel or carbon
of the top coat to the bond coat would be poor, due to steel substrate; t denotes the thickness of the substrate, T and N denote
the lack of mechanical bonding sites and very weak the specimen conditions.
 Y.C. Zhou et al. / Surface and Coatings Technology 157 (2002) 118–127 125

original substrate with thickness of 3.0 and 5.0 mm. On

the other hand, the following interesting phenomenon
can be observed: although the substrate thickness was
the same for three specimens tested, the characteristic
of crack density vs. strain was different for specimens
with different original substrate thickness.
In FGM specimens, no spallation of the top coat
occurred and only localized delamination was evident
along the interface between the FGM layers and the
substrate. As indicated in Fig. 13, no delamination was
observed between FGM layers, while multiple vertical
cracks were formed and extended near the interface.

4. Conclusions

Fig. 11. Plot of interface stress intensity vs. strain during four-point
bending test of non-FGM and FGM coating systems with stainless Uniaxial tensile and four-point bending tests of ther-
steel or carbon steel substrate. T denotes the specimen conditions. mal barrier ceramic coatings were conducted. The fol-
lowing conclusions were drawn from the fracture
The stress intensity factor can be obtained from Eq. characteristics observed:
(7), Eq. (8) and Eq. (12), using data from Fig. 9. The
stress intensity factors for our four-point bending exper- 1. For a thicker substrate, less strain is required to
iment are shown in Fig. 11. The stress intensity factor induce delamination near the top coatybond coat
shown in Fig. 11 for complete spallation is regarded as interface.
the fracture toughness of the interface. It is evident that 2. Vertical cracks first appeared on the surface of the
interface fracture toughness is Kcs4.26–7.21 MPa top coat layer, normal to the load axis.
m1y2 for specimens prepared under condition T, i.e. 3. When the surface crack tip was close to the top coaty
gun–substrate distance of 125–130 mm and gun oper-
bond coat interface, the crack appeared to kink from
ating power of approximately 38.5 kW. It is also shown
its original orientation. The kinked cracks propagate
that interface fracture toughness is greater for thicker
substrate material. On the other hand, interface fracture along the interface between the topybond coats.
toughness for specimens prepared under condition T is 4. Discontinuous delaminated cracks formed at the top
greater than that for specimens prepared under condition coatybond coat interface link-up, with eventual spal-
N. lation of the top coat ceramic layer.
It is known that interface strength and fracture tough- 5. Substrate thickness has an important effect on the
ness calculations for the shear lag and Suo–Hutchinson fracture characteristics of the TBC.
models are only valid for flat interfaces w12,19,23x. In
fact, the roughness of the bond coat has an important
effect on interface strength and interface fracture tough-
ness, as previously discussed. Therefore, a model that
considers the effect of bond coat roughness was pro-
posed by Evans and Hutchinson w24x to evaluate the
interface strength and interface fracture toughness.
In order to examine the effect of substrate thickness
on fracture characteristics, a substrate with sprayed bond
coat and top coats was cut to equal dimensions. The
plasma-sprayed substrates, with original substrate thick-
ness of 3.0 and 5.0 mm, were cut to 1.5-mm thickness.
The crack density vs. strain for these specimens is
shown in Fig. 12 and compared with data obtained from
a specimen with original substrate thickness of 1.5 mm.
Other conditions, such as bond and top coat thickness,
were the same. We note that the strain at which the first
crack took place was different for the specimens with Fig. 12. Crack density vs. strain in four-point bending test for speci-
the same thickness of substrate. The characteristic of mens with an original substrate thickness of 3.0 and 5.0 mm, but cut
crack density vs. strain, however, was the same for the to 1.5 mm in thickness.
126 Y.C. Zhou et al. / Surface and Coatings Technology 157 (2002) 118–127

Fig. 13. SEM cross-section of FGM specimens in four-point bending test: (a) center; (b) middle; and (c) edge of the specimen.

6. In testing of the FGM specimen, no spallation of the Acknowledgments

top coat was observed, and only localized delamina-
tion occurred along the interface between the FGM
The authors express their appreciation for financial
layers and the substrate. No delamination was evident
and logistic support, provided by Grant-in-Aid for COE
between FGM layers, although multiplicity of vertical
cracking did occur and extended toward the interface. (Center of Excellence) Research (No 11CE2003), and
7. Compared to non-FGM specimens, FGM samples for Scientific Research (B) (No 09555027), The Min-
exhibited enhanced interface strength. istry of Education, Science, Sports and Culture (Japan).
8. Interface fracture toughness, evaluated by the shear
lag model for the uniaxial tensile test, was 0.94 and Appendix A:
0.67 MPa m1y2 for type A and B coatings, respectively
(for specimens prepared under condition N). The
interface fracture toughness evaluated by the Suo– The energy release rate for interface cracking between
Hutchinson model for the uniaxial tensile test was two elastic layers according to Suo and Hutchinson w19x
Kcs1.05–1.27 and 1.0–1.17 MPa m1y2 for type A is given by:
and B coatings, respectively (for specimens prepared
c1 w P2 M2 PM z
under condition N). The results were very similar to Gs x q 3 q2 sing| (A1)
data obtained by the blister test method. 16 y Ah Ih yAIh2 ~
9. Interface fracture toughness evaluated by the Suo–
k1q1
Hutchinson model for the four-point bending test was where c1s with ks3y4n for plane strain and
Kcs4.26–7.21 MPa m1y2 for specimens prepared m1
under condition T. The interface fracture toughness 3yn
ks for plane stress and n is Poisson’s ratio, and
for specimens prepared under condition T is greater 1qn
than that for specimens prepared under condition N. m the shear modulus, respectively. Other non-dimension-
 Y.C. Zhou et al. / Surface and Coatings Technology 157 (2002) 118–127 127

al parameters A, I, angle g, S, h, a and b are defined References

as: w1x I. Yuri, T. Hisamatsu, K. Watanabe, Y. Etori, Trans. ASME J.
1 1 Eng. Gas Turbines Power 119 (1997) 506–511.
As Is w2x A. Rabiei, A.G. Evans, Acta Mater. 48 (2000) 3963–3976.
1qSŽ4hq6h2q3h3. 12Ž1qSh3. w3x Y.C. Zhou, T. Hashida, Int. J. Fatigue 24 (2002) 407–417.
w4x D.M. Zhu, R.A. Miller, J. Mater. Res. 14 (1999) 146–161.
1qa h w5x C.Y. Jian, T. Hashida, Compos. Eng. 5 (1995) 879–889.
sings6Sh2Ž1qh.yAI Ss hs (A2) w6x J. Cheng, E.H. Jordan, B. Barber, M. Gell, Acta Mater. 46
1ya H
(1998) 5839–5850.
GŽk2q1.yŽk1q1. GŽk2y1.yŽk1y1. w7x Y.C. Zhou, T. Hashida, Int. J. Solids Struct. 38 (2001)
as bs 4235–4264.
GŽk2q1.qŽk1q1. GŽk2q1.qŽk1q1. w8x G.C. Chang, W. Phucharoen, Surf. Coat. Technol. 32 (1987)
307–325.
where Gsm1 ym2, h is the thickness of the coating, H w9x G.C. Chang, W. Phucharoen, Surf. Coat. Technol. 30 (1987)
is the thickness of the substrate, and subscripts 1 and 2 13–28.
refer to the coating and substrate, respectively. The load w10x J.W. Hutchinson, Z. Suo, Adv. Appl. Mech. 29 (1992) 63–191.
w11x A.G. Evans, M.Y. He, J.W. Hutchinson, Acta Mater. 45 (1997)
parameters P and M are given by: 3543–3554.
w12x M.S. Hu, A.G. Evans, Acta Metall. 37 (1989) 917–925.
M3 w13x M. Yanaka, Y. Tsukahara, N. Nakaso, N. Takeda, J. Mater. Sci.
PsP1yC1P3yC2 MsM1yC3M3 (A3)
h 33 (1998) 2111–2119.
w14x Y.P. Jiang, Y.C. Zhou, L. Xiao, Y. Pan, Z.L. Long, L.Q. Zhou,
with Residual stress and stress–strain properties of electrodeposited
nickel coating, Surf. Coat. Technol., submitted for publication.
S SB1 1E S w15x Y.J. Kim, T. Hashida, H. Takahashi, K. Fujii, M. Kanazawa,
C 1s C 2s C yDq F C 3s (A4) Spray 33 (1996) 1–8, In Japanese.
A0 I0 D h 2G 12I0 w16x Y.C. Zhou, T. Hashida, Int. J. Solids Struct. 38 (24–25) (2001)
4235–4264.
1 w17x Y.C. Zhou, T. Hashida, in: R. Townsend, et al. (Eds.), Life
A0s qS Assessment of Hot Section Gas Turbine Components, Cam-
h
bridge University Press, 2000, pp. 155–172.
1 TS w B 1 E2 B 1E z D
I0s UTSx3CDy F y3CDy Fq1|q3
w18x Y.C. Zhou, T. Hashida, C.Y. Jian, J. Eng. Mater. Technol.
3V y D hG D hG ~ h ASME, in press.
w19x Z. Suo, J.W. Hutchinson, Int. J. Fracture 43 (1990) 1–18.
B 1 E 1T
W
CDy Fq 3 XT (A5) w20x G.J. Petrus, B.L. Ferguson, J. Therm. Spray Technol. 6 (1997)
D hG h Y 29–34.
w21x G.C. Chang, W. Phucharoen, R.A. Miller, Surf. Coat. Technol.
1q2ShqSh2 30 (1987) 13–28.
Ds (A6) w22x G.C. Chang, W. Phucharoen, R.A. Miller, Surf. Coat. Technol.
2hŽ1qSh. 32 (1987) 307–325.
w23x S.J. Howard, Y.C. Tsui, T.W. Clyne, Acta Metall. Mater. 42
where P1 and P3 are loads per unit thickness, and M1 (8) (1994) 2823–2836.
and M3 are moments per unit thickness, as defined in w24x A.G. Evans, J.W. Hutchinson, Acta Metall. 37 (3) (1989)
w19x. 909–916.