JTTEE 3:184-190

'9 International

A Test for Coating Adhesion on Flat

Substrates--A Technical Note
S.H. Leigh and C.C. Berndt

The standard tensile adhesion test (TAT), ASTM C633, has been modified to perform multiple tests on
flat and wide substrates. The TAT geometry, which specifies a 1-in. (25.4-mm)-diameter cylindrical sub-
strate, has been used as the pull-off bar. Two renditions of this test were implemented and the Weibull
moduli and characteristic stresses for both test methods obtained. The modified TAT, termed as the sin-
gle-bar (SB) method, yields a higher Weibuii modulus and characteristic strength than the other method,
which is termed the double-bar (DB) method. It is believed that the different test results between the two
methods arise from different stress distributions near the interface of the coating and substrate.

1. Introduction In the present work, TATs were performed to evaluate the ad-
hesion strength of thermal spray coatings on flat substrates, i.e.,
THE adhesion strength between the coating and the substrate is substrates that need not be of cylindrical geometry or metal.
one of the most important properties from which the lifetime 1-3 This test configuration is compared with another tensile adhe-
and the quality of a specific application can be estimated. Nu- sion test method, which is termed as the double-bar (DB)
merous tests can be used to evaluate the adhesion strength of method, and the test results are compared. A Weibull analysis,
thermal spray coatings. 4-7 The tensile adhesion test (TAT)8 is which is based on the "weak link theory," is used to analyze the
relatively simple to perform and is a widely used method in in- TAT data.
dustry and scientific laboratories. However, there are some
shortcomings, 3-5'9-11and many controversies still exist. 5,7,11,12
The TAT is often used as a quality control tool to determine
2. Modified TAT Methods
spraying conditions, surface conditions of substrate, grit blast-
The modified configuration of the TAT, designated as the SB
ing conditions, coating thickness, etc.
method, has been designed for self-alignment and to accommo-
The TAT has been standardized and there are four main date flat and wide thermal spray-coated specimens. The basic
standards: ASTM C633-79 (USA), 8 DIN 50 160-A (Ger- configuration of the SB method is similar to the Elcometer adhe-
many),13 AFNOR NF A91-202-79 (France),14 and JIS H8666- sion test (Elcometer Instruments Limited, Droylsden, England),
80 (Japan). 15The test configuration of ASTM C633 is given by which has been used to determine the adhesion of paint. Figure
Fig. 1. Other standards have the same basic configuration; how- 2 represents the SB method, which can accommodate speci-
ever, the substrate dimensions vary from 25.4 to 40 mm in di- mens up to 6 • 6 x 6 in. (152 x 152 x 152 mm). The specimen is
ameter.16'17This difference in diameter results in a variation of positioned between two plates that are 7 x 7 in. (178 x 178 mm)
the stress singularity near the edge of the pull-off bar, i.e., the wide. The bottom plate has a 1-in. (25.4-mm)-diameter hole at
stress distribution in the vicinity of the pull-off bar is not uni- the center, through which the glued pull-off bar passes and pro-
form; therefore, the adhesion value and its variation may change vides tensile stress to the specimen. This bottom plate and the
between coatings. universal joints at the top and bottom of the fixture result in a
Another distinction of the above standards is that the size, self-aligning TAT system. This geometry also has an advantage
shape, and material of the substrate are restricted (e.g., 25.4- in that several tests can be performed on each flat substrate,
mm-diameter cylindrical shape of metal for ASTM C633 com- since the pull-off bar can be repositioned for each TAT.
pared to 40 mm for the DIN standard). This limits some The DB method is shown in Fig. 3. This method requires no
applications of the TAT to geometries that are determined by special testing fixture except two pull-offbars and two universal
standards rather than by the complex geometries of substrates joints, which are attached to both sides of the flat specimen. Fig-
that are currently being spray-coated. As thermal spray technol- ure 4 shows the simple fixture required to align the two pull-off
ogy covers more fields of application, it is expected that there bars during the epoxy curing process. This configuration also al-
will be more need to utilize a variety of substrates, such as ce- lows multiple tests on flat geometries, but at the expense of
ramics and materials of complex shape. Some modifications, maintaining alignment of the two glued attachment bars.
therefore, need to be made to enable adhesion-type tests on such
substrates.
3. Materials and Experimental Procedure
I Keywords: ASTMC633, tensile adhesiontest, Weibull analysis, sin- Alumina powder (Metco 101) was sprayed on top of the bond
gle-bar, double-bar,stress distribution coat (NiCrAIY, Amdry 961) onto 6 x 6 x 0.3 in. (152 x 152 x 7
S.H. Leigh and C.C. Berndt, Thermal Spray Laboratory, Department ram) flat substrates. The substrates were grit blasted before
of Materials Science and Engineering, State University of New York at spraying using 24-mesh alumina at an air pressure of approxi-
Stony Brook, Stony Brook, NY 11794-2275. mately 60 psi (0.4 MPa).

184--Volume 3(2) June 1994 Journal of Thermal Spray Technology

 7 in. (180 m m )

6 in. (152 m m )

1 in. (25.4 m m )
I I
L---- -1
' ' I
universal
,I I,
I
I
I
I
I
II O II
joint I
tL---
_1
-I

I-
tf -1L 7 SUBSTRAT|

SPRAYED
COATING

EPOXY
substrate I I
kk / ) PULL-OFF-B,*r~ , ,,,- I--,,.'T ,,,,,,;
epoxy V
Fig. 2 Schematic of TAT assembly for single-bar (SB) method.

1 in. / 2 5 . 4 m m ~
PULL-OFF
A
coating / k SUBSTRA1

(I 5I

SPRAYED
COATING

C q EPOXY
1 in. (25.4 mm)
Fig. 3 Schematic of TAT assembly for double-bar (DB) method.
universal
o1:
f I-'- -- -- -- ~tqI
joint II
Westbury, NY, USA) (Table 1). The spray condition was not op-
timized for adhesion strength, since it was not intended that this
II work focus on the magnitude of adhesion strength but rather on
the testing method. The thickness of alumina coating was be-
tween 0.006 and 0.008 in. (0.15 and 0.20 mm), and the bond coat
I ,r---I
' ,' I was between 0.002 and 0.004 in. (0.05 to 0.10 mm).
Tensile adhesion tests were performed using a hydraulic In-
stron (Model AW2414-1, Instron Corporation, Canton, Massa-
chusetts). The capacity of the load cell was 200 kN. Calibrations
Fig. 1 Configuration of tensile adhesion test specified by ASTM of the load cell were performed automatically by the controller
C633. attached to the Instron. The speed of the crosshead was 0.030
in./min (0.013 mm/sec); the test conditions are summarized in
Table 2. The fracture stress (average stress) was calculated by
The plasma spraying process was performed under atmospheric dividing the failure load by the testing area [0.7854 in.2 (506.7
conditions using a Metco 3 MB gun (Perkin Elmer, Metco Division, mm2)].

Journal of Thermal Spray Technology Volume 3(2) June 1994---185

 9 in. (228.6 mm) 8 r , '

8 in. (203.2 ram)

1 in. (25.4 ram)
Total n u m b e r of tests:20
I I I I
I I p<,,~, ~,
O0
h3
[ ' , E-~ IKX~
,'./ v \~
I I I I

0
(I ' ' U ix.

r..4
~

X
.N

:~

rvH I,,< x ~

Fig. 4 Fixture for alignment of double-bar (DB) test specimen.

Z
Table 1 Spray conditions for air-plasma spraying of
alumina powder

Gun Metco 3MB

Nozzle GH 0.0 0.5 1.0 1.5 2.0 2.5 3.0 5.5 4.0 4.5 5.0
Voltage, volts 60
Current, a~hpems 500
Argon flow rate, SLPM 85
ADHESION STRENGTH (i?iPa)
Hydrogen flow rate, SLPM 15 (a)
Powder carrier gas (Ar) flow rate, SLPM 38
Powder feed rate, g/min 50
Spray distance, in. (cm) 3 (8) i [ F 1-

Material and average particle size A1203 (Metco 101), 53 t.trn

7
Total number of tests=23
Table 2 Summary of tensile adhesion test conditions and
materials oo 6
oo
Test instrument and Model No. Hydraulic Instron (Instron
5
Corporation,Canton, MA), AW2414-1
Load cell capacity 200kN
Crosshead speed 0.030 in./min (0.013 mm/sec) O 4
Diameter ofpull-offbar I in. (25.4mm)
Length of pull-off bar 1.2 in. (30.5 mm) L~
Adhesive bonding agent (Epoxy) FM-1000 film adhesive (American 3
Cyanamid Company)
Adhesive curing temperature 170 ~ + 6 ~ (340 ~ _+10 ~ Z 2
Adhesive curing time 60 min
Test materials Alumina coatings on flat substrates
[6 x 6 x0.3in.(152 x 152 x 7 mm)l
Number of test specimens i. 20 for single-bar (SB) method
0
00 0.5 1.0 15 2.0 2.5 3.0 3.5 4.0 4.5 5.0
4. Results and Discussion
ADHESION STRENGTH (MPa)
Figure 5 shows test result histograms of the two TAT meth-
(b)
ods. The adhesion strengths as measured by the DB method are Fig.5 Adhesion strength vs number of tests for (a) single-bar and (b)
more widely distributed than those of the SB method, shown in double-bar methods.
Fig. 6. The arithmetic mean of adhesion strengths measured by
the SB method is about 38% higher than the DB method; the SB
method yields a higher median and a lower standard deviation. The Weibull distribution can be expressed in the linear form
as"
4.1 Weibull Analysis of TAT Data
The adhesion test result of thermal spray coatings has a InIln(ll~_pll=mln(~-Gu,-mlnoo+ InV (I)
large variability, which makes the interpretation of data and
interlaboratory comparison difficult. A Weibull analysis is where P is fracture probability for the stress c, Ou is the stress
one method of finding a parameter to assess the variability below which fracture is assumed to have zero probability, ao is
and to index the coating adhesion strength of thermal spray a characteristic stress at which 63.2% of the specimens fail, m is
coatings. 18,19 the Weibull modulus, and V is the v o l u m e of the specimen.

186----Volume 3(2) June 1994 Journal of Thermal Spray Technology

 I I I I i

/, (Ioul)h~ b a r m e t h o d
45 I 9 single bar method

4 C

3.5 m=16

3 2

2.5

2
-4 /~ / 9
15 /
5 __~_H._ i /i i i i i
20 i 5 i0 05 oo 05 10 15 20
1
In(failure load), MPa
0.5
Fig. 7 Weibull plot for two test methods.
0

mean median standard maximum minimuln

deviation
Fig. 6 Summary of tensile adhesion test data. Table 3 W e i b u l l statistics o f t w o a d h e s i o n test m e t h o d s

Characteristic
Test method Weibull modulus, m strength, Oo, MPa
The estimator chosen to determine the value of P for thejth Single-bar(SB) method 3.1 2.6
strength 2~ was: Double-bar(DB) method 1.6 1.9
j-0.5
PJ - n (2) Hopman's analysis can still be used for a qualitative explanation
of the current problem. The applied stress is concentrated near
where n is the batch size and Pj is the fracture probability (P) for the edge of the pull-off bar, i.e., the singular stress exists in the
thejth strength. vicinity of the pull-off bar. As the substrate thickness increases,
All of the coating and substrate materials were constant in the stress values in the substrate and the pull-off bar decrease;
this study. Therefore, a higher value of rn indicates that the test therefore, the failure stress increases (Table 4). Note in Fig. 8
method is a more reliable indicator of failure behavior and the that a stress field is extended to the bottom side of the substrate.
material property of strength. In the case where another bar is attached to this side of the sub-
The Weibull moduli (m) and the characteristic stresses (fro) for strate (i.e., the configuration of the DB method), this bar would
the two test methods ~re obtained from the Weibull plot (Fig. 7) and also produce a stress field that would be expected to influence
are listed in Table 3. The Weibull modulus is obtained from the the overall stress field near the coating/substrate interface. Thus,
slope and the characteristic stress from the intercept on the prob- the DB method yields lower average failure strengths and char-
ability axis of the Weibull plot. The characteristic stress is defined acteristic strengths than the SB method. Equation 1 is based on
as the stress below which 63.2% of the specimens fail and can be the assumption that the tensile stress is uniform over the entire
considered a meaningful parameter to index the adhesion strength
volume of the specimen; 2~ therefore, the differences in Weibull
of coatings. Both the Weibull modulus and the characteristic stress
moduli probably arise from changes in the stress distribution over
of the SB method are higher than those of the DB test. Berndt 10cal-
the coating/substrate interface.
culated the Weibull modulus of plasma-sprayed alumina coatings
Figure 8 shows that the stress field at the bottom of the sub-
using the TAT data reported by Hermanek. 2 1 These values range
strate decreases as the thickness of the substrate increases. Thus,
from 1.4 to 3.8 and can be compared with those obtained in this
study, which have quite similar results, i.e., 3.1 and 1.6 for the SB it is expected that if the substrate is sufficiently thick [about 1 to
and DB methods, respectively. 1.5 in. (2.54 to 3.81 mm)], then the test results of both methods
would be similar. The tables in Fig. 8 indicate that the maximum
stress also decreases as the thickness of the substrate increases;
4.2 Stress Distribution near Edge o f the Pull-off B a r
thus, thicker substrates yield higher failure stresses. In addition,
The TAT is based on the assumption that a uniform stress dis- longer pull-off bars, which produce more uniform stress distri-
tribution exists over the entire testing area. The adhesion bution near the coating/substrate interface, may reduce the dif-
strength is calculated by dividing failure load by testing area, ference between the two test results. 23
and the "average stress" at which failure occurs is measured.
Thus, stress singularities22-26near the pull-off bar that influence
4 . 3 Comments on the Two Adhesion Test Methods
the test result are disregarded.
Hopman 27 investigated the stress distribution near the pull- Two kinds of tests, SB and DB, based on ASTM C633, were per-
off bar edge of four substrate thicknesses subjected to the SB formed. Both of the test methods are direct pull-off tests that incor-
method using finite element analysis (FEA), as shown in Fig. 8. porate various test fixtures. One advantage of these test methods is
Although the quantitative stress values and the location of the iso- that coatings on flat substrates can be tested without preparing them
stress line are most probably different from the present work (since in the form of a standardized tensile adhesion test specimen (e.g., 1-
the sample shape and the materials of the test assembly are different), in.-diameter bar) and various materials can be used. For example,

Journal of Thermal Spray Technology Volume 3(2) June 1994----187

 CONTOURLINE VALUES
(No.) (MPa)
1 19
2 44
3 68
4 92
5 117
/ 9 ~=~ 6 141
/ / ._3~ 7 166
/ / ,%~ 8 ,90
/ / , [A.[i~' 19 239 ~ 8 9 1 0
/ /--J ,,.%.--..-'~_/
~ /.,. - 6 ~ 5

CONTOURLINE VALUES
I (No)
1
(MPa)
27
2 39
3 52
4 64
5 77
6 90
7 102
8 115
9 127
10 140

8 10 6
10 9

1X1"~,.__ 2 CONTOURLINE VALUES

(No.) (MPa)
1 11 CONTOURLINE VALUES
2 19 (No.) (MPa)

1 4 34 2 10
2mm 5 42 4mm 3 15
6 50 4 20
7 58 5 25
8 65 6 30
i 9 73 7 35
10 80 8 45
2~ 2 9 46

Fig. 8 Contour plot of equivalent stresses near the pull-off bar edge for four substrate thicknesses (from Ref 27). All pull-off bar sizes are the same.

ceramics or materials that are difficult to machine can easily be ings sprayed onto different sized substrates can also be different.
adopted as substrates. For instance, the specimen size may induce different residual
The sample preparation of the SB method is comparatively stresses that influence the adhesion strength of the coating.
simple. The DB method has no specific test fixture except the Therefore, the larger size of specimens can approach the proper-
two bars on both sides of the specimen, although a special fix- ties of the product in service. This is one advantage of perform-
ture is required for sample preparation of the DB method to ing adhesion tests directly on flat and wide substrates.
maintain alignment (Fig. 4). In addition, the DB method cannot
be used unless both sides of the substrate are flat and reasonably
parallel to the coated surface, while the SB method can be used 5. C o n c l u s i o n
on a specimen that has a flat coated surface.
Every standard adhesion test (ASTM, DIN, AFNOR, JIS, The TAT is relatively simple to perform and is a widely used
etc.) specifies the specimen size which, in most cases, is less adhesion test method. Although there are some inherent short-
than the size of coatings used in service. The properties of coat- comings, it can be modified and developed to establish a more

188---Volume 3(2) June 1994 Journal of Thermal Spray Technology

Table 4 Measured stresses and forces at failure of paint coat and calculated m a x i m u m stresses (from Ref. 27)

Measured Calculated
Substrate Mean stress at Maximum stress Maximum
thickness, failure, Force at failure, in substrate, stress in dolly,
mm MPa N MPa MPa
0.5 2.0 157.0 238 214
1 2.9 216.9 140 102
2 4.2 334.6 80 65
4 7.0 549.8 50 41

suitable testing procedure, since it is still a popular method in in- Coatings on WC-Co Substrates, Thin Solid Films, Vol 154, 1987, p
dustry for quality control and in the laboratory environment for 361-375
optimization of spray conditions. 10. C.C. Berndt, Tensile Adhesion Testing Methodology for Thermally
A modified TAT, based on ASTM C633 and termed the SB Sprayed Coatings, J. Mater. Eng., Vol 12, 1990, p 151-158
method, was devised to accommodate flat and wide substrates. 11. C.C. Berndt and R. McPherson, A Fracture Mechanics Approach to the
The test results were compared with another TAT method, Adhesion of Flame and Plasma Sprayed Coatings, Trans. Institute of
Engineers, Vol 6, No. 4, 1981, p 53-58
termed the DB method. A Weibull analysis was used to analyze
12. C.C. Berndt, The Need for Standardization of Testing Techniques in the
and compare the two TAT methods. The Weibull modulus (m)
Thermal Spray Industry, in Thermal Spray Research and Applica-
and characteristic stress (~o) can be used to characterize the
tions, T.F. Bernecki, Ed., ASM International, Materials Park, OH,
coating integrity and/or coating adhesion. The SB method has a
1991, p 325-330
higher characteristic stress and Weibull modulus than the DB
13. DIN50 160, Determination ofAdhesive Strength in the Traction-Adhe-
method. The difference in the test results seems to be derived sive Strength Test, DIN-Normen, Beuth Verlag Gmbht, Berlin 30, Aug
from changes in the stress distribution near the interface of the 1981
coating and the substrate. Such measurement differences would 14. AFNOR NF A91-202-79, Characteristics and Methods of Test for Met-
be reduced by using thicker substrates and/or longer pull-off al Spraying, Association Francaise deNormalisation, Tour Europe,
bars. Cedex 792880, Paris, Oct 1979
15. JIS H 8666-80, Thermal Sprayed Ceramic Coatings, Japanese
Standards Association, 1-24, Akasaka 4, Minato-ku, Tokyo, Ja-
Acknowledgments pan, 1981
This work has been partially supported by the STRATMAN 16. Y. Shimizu, M. Sato, M. Kobayashi, and K. Maeda, Effect of Test
program of the National Science Foundation under grant num- Specimen Size upon Adhesive Strength of Flame Sprayed Coatings, in
ber DDM9215846. Thermal Spray Coatings: Properties, Processes and Applications,
T.E Bernecki, Ed., ASM International, Materials Park, OH, 1992, p
257-262
References 17. O. Ambroz and J. Krejcova, Determination of the Adhesive and Cohe-
1. H. Wang and W. Montasser, Degradation of Bond Coat Strength under sive Fracture Modes of the Adhesion Tensile Test, in Thermal Spray:
Thermal Cycling-Technical Note, J. Thermal Spray Technol., Vol 2, International Advances in Coatings Technology, C.C. Berndt, Ed.,
No. 1, 1993, p 31-34 ASM International, Materials Park, OH, 1992, p 921-927
2. H. Wang, H. Herman, G.A. Bancke, and M. Wood, Flame Rig Testing of 18. A. De S. Jayatilaka and K. Trustrum, Statistical Approach to Brittle
Thick Thermal Barrier Coatings, in Protective Coatings: Processing Fracture, J. Mater. Sci., Vol 12, 1977, p 1426-1430
and Characterization, R.M. Tazici, Ed., TMS, Warrendale, PA, 1990, 19. S.J. Grisaffe, Analysis of Shear Bond Strength of Plasma-Sprayed Alu-
p 155-163 mina Coatings on Stainless Steel, NASA TN D-3113, NASA-Lewis Re-
3. C.C. Berndt and R.A. Miller, Mechanical Property Measurements of search Center, Cleveland, OH, USA, July 16, 1965
Plasma-Sprayed Thermal-Barrier Coatings Subjected to Oxidation, 20. B. Bergman, On the Estimation of the Weibull Modulus, J. Mater. Sci.
Ceram. Eng. Sci. Proc., Vol 6, Nos. 7,8, 1984, p 479-490 Lett., Vol 3, 1984, p 689-692
4. P.R.Chalker, S.J. Bull, and D.S. Rickerby, A Review of the Methods for 21. F.J. Hermanek, Determining the Adhesive/Cohesive Strength of Thin
the Evaluation of Coating-Substrate Adhesion, Mater Sci. Eng., Vol Thermally Sprayed Deposits, Weld. J., Vo157, 1978, p 31-35
140A, 1991, p 583-592 22. Y. Inoue, A. Noutomi, M. Toyoda, and A. Izuha, Applicability of Stress
5. S.D. Brown, B.A. Chapman, and G.P. Wirtz, Fracture Kinetics and the Singularity Parameter for Evaluating Bond Strength of Plasma Sprayed
Mechanical Measurement of Adherence, in Thermal Spray Technol- Coatings, in Thermal Spray Coatings: Properties, Processes and
ogy: New Ideas and Progresses, D.L. Houck, Ed., ASM International, Applications, T.E Bernecki, Ed., ASM International, Materials Park,
Materials Park, OH, 1988, p 147-157 OH, 1992, p 251-255
6. D.S. Rickerby, A Review of the Methods for the Measurement of Coat- 23. W. Han, E.E Rybicki, and J.R. Shadley, An Improved Specimen Ge-
ing Substrate Adhesion, Sulf. Coat. Technol., Vo136, 1988, p 541-557 ometry for ASTM C633-79 to Estimate Bond Strengths of Thermal
7. S.H. Leigh, Masters thesis, State University of New York at Stony Spray Coatings, J. Thermal Spray Technol., Vol 2, No. 2, 1993, p 145-
Brook, Stony Brook, New York, May 1993 t50
8. ASTM C633-79, Standard Test Method for Adhesion or Cohesion 24. M.L. Williams, Stress Singularities Resulting from Various Boundary
Strength of Flame Sprayed Coatings, American Society for Testing and Conditions in Angular Corners of Plates in Extension, J. Appl. Mech.,
Materials, Philadelphia, 1979 Trans. ASME, Vo119, No. 74, 1952, p 526-528
9. P.C. Jindal, D.T. Quinto, and G.J. Wolfe, Adhesion Measurement of 25. V.L. Hein and E Erdogan, Stress Singularities in a Two-Material
Chemically Vapor Deposited and Physically Vapor Deposited Hard Wedge, Int. J. Frac. Mech., Vol 7, No. 3, 197 !, p 317-330

Journal of Thermal Spray Technology Volume 3(2) June 1994--189

26. D.B. Bogy, Edge-Bonded Dissimilar OrthogonalElastic Wedges Under 27. EC. Hopman, The Direct Pull-Off Test, JOCCA, Vol 7, 1984, p 179-
Normal and Shear Loading, J. Appl. Mech., Trans.ASME, Vo135, No. 184
91, Series E, 1969, p 460-466

190---Volume 3(2) June 1994 Journal of Thermal Spray Technology