JTTEE5 2:145-150

9 International

An Improved Specimen Geometry for ASTM

C633-79 to Estimate Bond Strengths
of Thermal Spray Coatings
W. Plan, E.E Rybicki, and J.R. Shadley

ASTM Standard C633-79, "Standard Test Method for Adhesion or Cohesive Strength of Flame-Sprayed
Coatings," is widely used in industry and research for evaluating bond strengths of thermal spray coat-
ings. Tests are conducted by applying the coating to the end of a cylindrical test specimen 25.4 mm (1 in.)
in diameter by 25.4 m m (1 in.) long. The coating surface is bonded to an uncoated cylinder of the same
material and geometry. The force required to pull the cylinders apart is recorded. The bond strength is
calculated by dividing the maximum force by the area of the l-in. diameter cylinder assuming that the
stress is uniform over the area where the debonding occurs.
A combination of finite-element stress analysis and experimental stress analysis using strain gages was
used to evaluate the stresses at the interface between the coating and substrate. Finite-element analysis
of the standard specimen geometry shows that the m a x i m u m stress at the coating interface can be 25 %
higher than the average stress. An elongated specimen was selected, constructed, and evaluated to pro-
duce the uniform stress distribution assumed by ASTM C633-79. Strain gage measurements and epoxy
tensile tests have verified that the bond strengths measured with the elongated specimen provide better
estimates of bond strengths than tests with the standard specimen.

1. Introduction characterize the bond strength of thermal spray coatings, al-

though more research is needed on specimen preparation, test
BOND strength is an important property of thermal spray coat- procedure, data repeatability, failure mode, and debond mecha-
nisms.t2,10-14/
ings. High bond strength of a coating is associated with higher
erosion, corrosion, and abrasion resistance. There are many test- Following ASTM C633-79, tests are performed by applying
ing standards to evaluate the bond strength of thermal spray the coating to the end of a cylindrical test specimen 25.4 m m ( 1
coatings. Currently, four major standards are used in industry in.) in diameter by 25.4 mm (1 in.) long. The coated surface is
and research. These are ASTM Standard C633-79 (USA), DIN bonded to another cylinder of the same geometry. The force re-
50 160-A (Germany), AFNOR NF A91-202-79 (France), and quired to pull the cylinders apart is recorded. As shown in Fig. 1,
JIS H8666-80 (Japan). [1-3IA,-~
~ a n o u g h each standard uses a differ- bonded specimens are mounted in a self-aligning device. The
ent specimen geometry, test method, and analysis procedure, the geometry of the ASTM C633-79 standard specimen is shown in
Fig. 2. The bond strength or ultimate bond stress is calculated by
primary goal of each standard is to determine the degree of ad-
dividing the ultimate force by the cross-sectional area of the l-
hesive or cohesive strength between the coating and the sub-
in. diameter cylinder. This procedure inherently assumes that
strate. ASTM Standard C633-79, "Standard Test Method for
the stress is uniform over the cross-sectional area where the
Adhesion or Cohesive Strength of Flame-Sprayed Coatings," is
coating is bonded. Although this assumption is convenient,
widely used in the United States for evaluating bond strengths of
there are several characteristics of the specimen design, includ-
thermal spray coatings. The application of ASTM Standard
ing the geometry of the specimen and the mismatch of material
C633-79 can be traced to 1959.[41This standard is currently used
stress-strain constants between the coating and substrate, that
to test thermal spray coatings, thin fihns, epoxies, and adhesives
can lead to a nonuniform stress distribution at the bonding inter-
in the aerospace industry, automobile industry, oil and gas in-
face. A nonuniform stress distribution infers that there is a place
dustry, chemical industry, medical industry, and lbod process- on the coating where the stress is higher than the average stress
ing industry. [5-91 Meanwhile, much research has been done on calculated by the applied force divided by the area. To better in-
the test method, data analysis, failure mode, and debond mecha- terpret the results of ASTM C633-79, it is important to have an
nisms. Research results indicate that ASTM C633-79, as well as estimate of the stress distribution applied to the coating during
the other standards mentioned above, will continue to be used to the test. The objective of this study is to examine the ASTM
C633-79 test procedure for accuracy and develop improve-
ments in the procedure to increase the accuracy.
Keywords: ASTM C633-79, bond strength, debond stress, finite ele-

I ment stress analysis, mechanical properties, strain gauge

measurements, stress distributions, test specimen geome-
try, thermal spray coating, ultimate bond stress

W. Hart, E.F. Rybicki, and J.R. Shadley, Mechanical Engineering De~

partment, The University of Tulsa, Tulsa. Oklahoma
2. Goals and Approach
The goals of this work are to (1) evaluate the stress distribu-
tion in a test specimen at the interface between the coating and

Journal of Thermal Spray Technology Volume 2(2) June 1993---145

 Force Threaded Hole

Bolts
Coating/Epoxy i
]_ 25.4 mm ~]

Epoxy

Coating
Section A-A
Specimer Fig. 2 ASTMC633-79 standard specimengeometry.

Bolts 3. Stress Distribution at the Interface

Between the Coating and Substrate
A finite-element model was developed to examine the stress
distribution at the interface between the coating and the sub-
strate. Figure 3 shows the two standard specimen halves bonded
together with epoxy according to the test procedure and the axi-
symmetric finite-element model of this structure. The focus of
this work is on the interface between the epoxy and the sub-
Force strate. To provide accurate information on how the stress
changes in the epoxy and in the substrate nearest the inter-
Fig. 1 ASTM C633-79 self-aligningtest fixture. face,the finite-element grid in these regions is more refined than
in other regions of the finite-element model. Accordingly, the
0.508-ram (0.02-in.) thick layer of the epoxy was divided into
the substrate and (2) design a new specimen to have a uniform four finite-element layers of thickness 0.127 mm (0.005 in.). A
stress distribution at the interface, as assumed by ASTM C633- tensile load of 13.34 kN (3000 lb) was applied to the model, as
79. The approach to achieving these goals is defined by the fol- shown in Fig. 3.
lowing steps: To verify the finite-element model, eight strain rosettes were
i. Develop a finite-elementmodel tbr the standard specimen installed on the surface of a standard specimen. Four strain ro-
settes were located on the surface near the bonding interface.
2. Apply a tensile load and evaluate the stress distribution at the in-
terface between the coating and substrate The other four strain rosettes were placed at locations where the
finite-element analysis indicated that high stresses occur (see
3. Verifythe finite-element model by experimental stress analysis
Fig. 4). Strains then were measured when a tensile load of 13.34
4. Compare the finite-element analysis predictions of the stress dis- kN (3000 lb) was applied to the standard specimen using a ten-
tribution at the debonding region with the uniform stress distribu- sile tester. Comparisons of surface stresses on the specimen, ob-
tion assumed by ASTM C633-79 tained from the strain gages, with the finite-element analysis
5. Develop a finite-element model for an elongated specimen that surface stresses are shown in Fig. 4. As shown, the stresses cal-
provides a uniform stress distribution at the interfacial debonding culated from the gage measurements are in good agreement with
region; verify the model with experimental stress analysis the finite-element analysis stresses. This provides confidence
6. Conduct debond tests for both the standard and elongated speci- that the finite-element model is valid and that stress distribu-
mens and compare the debond strengths obtained tions predicted by the model are accurate.
Step 4 of the approach is important because a nonuniform in- The axial stress distribution predicted by the model at the in-
terface stress distribution can lead to misinterpretation of the terface of the "standard" specimen is shown in Fig. 5. The stress
test results. distribution at the interface of the standard specimen was found

146----Volume 2(2) June 1993 Journal of Thermal Spray Technology

 C-,en~r Urw To alter the stress distribution at the interface, the geometry
of the specimen was changed. The specimen was elongated to
provide stiffness to the bonded interface area near the cavity. A
finite-element model of the elongated specimen was developed.
The elongated specimen was identical to the standard specimen
in all aspects, except that the length was increased from 25.4 to
38.1 mm (1 to 1.5 in.). Again, a tensile load of 13.34 kN (3000
lb) was applied to the finite-element model. To validate the fi-
E nite-element model for the elongated specimen, eight strain ro-
E
settes were installed on the elongated specimen surface. Strains
04 were measured when a tensile load of 13.34 kN (3000 lb) was
applied. Comparisons were made between the surface stresses
on the specimen and the surface stresses predicted by the finite-
E element model. As shown in Fig. 6, the stresses calculated from
E
strain gage measurements agreed with the finite-element model
04
predictions.
The stress distribution at the interface, computed from the fi-
nite-element model, is shown in Fig. 5 by the line corresponding
to the "elongated" specimen. This figure shows that the axial
stress distribution at the interface of the elongated specimen is
nearly uniform and almost identical to the average stress of 26.2
MPa (3800 psi) assumed by the standard analysis procedure.
Force Cenlor Line
This is an improvement over the standard specimen, for which a
(a) Specimen Assembly (b) AxisymmelrJc Grid nonuniform stress distribution was found.
~Bm,,. R

5. Experimental Bond Strength

Comparisons of Standard and
Elongated Specimens Using Epoxy
(c) Grid For Area A Three standard specimens and three elongated specimens
made of 1018 carbon steel were bonded together using a 3M
epoxy designated EC-1386. Following the procedure given in
Fig. 3 Finite-element model for standard specimen.
ASTM C633-79, alignment was provided by the double swivel
self-aligning fixture. Verification of the alignment for the self-
aligning fixture was conducted using three axial strain gages
to be nonuniform. The peak stress is almost 34.47 MPa (5000 that were equally spaced around the specimen circumference. A
psi), which is 25% higher than the average stress of 26.2 MPa load was applied using the fixture and a material testing system
(3800 psi) assumed by the A S T M C633-79 analysis procedure. MTS 810. Each gage reading was within 1.5% of the average
This implies that bond strengths determined using the standard value of the three gages.
specimen and procedure could be significantly underrated due To determine bond strengths, tensile loads were applied to
to the existence of the peak stress at the outer edge of the speci- the specimens until debonding occurred and ultimate loads were
recorded. The tensile bond strength of the epoxy to the substrate
men that is higher than the value assumed by the ASTM stand-
was calculated using the ultimate force divided by the cross-sec-
ard
tional area. Bond strengths of the epoxy, as determined by this
procedure for both the standard and elongated specimens, are
4. Modification to the Standard Specimen shown in Fig. 7. For the same epoxy and the same test proce-
dure, the bond strengths using the elongated specimen are
Geometry higher than the bond strengths of the standard specimea. This re-
sult implies that bond strengths measured using the standard
As shown in Fig. 2, there is a cavity in the standard specimen
specimen can be significantly lower than bond strengths deter-
where the specimen is drilled for tapping. The distance from the mined using the elongated specimen.
tip of the cavity to the surface where the epoxy is applied is only
6.35 mm (0.25 in.). The cavity makes the specimen less stiff at
the center near the cavity than it is at the outer edge. Tensile 6. Discussion and Conclusions
stresses are unable to distribute evenly across the interface.
Thus, because of the close proximity of the cavity to the inter- Finite-element analysis has been found to be a useful tool for
face, stresses are not as high near the center of the specimen as evaluating the stress distribution at the interface between coat-
they are at the outer edge. ings and substrates. Strain gages mounted on the surface of

Journal of Thermal Spray Technology Volume 2(2) June 1993---147

 Applied Load: 3000 Ib
60
9.R 4 m m
FE Axial Stress
50 For 3rce
FE Hoop Stress
40 $traln Gauges
CL
30
V

U)
U)
(9 20
r

or)
10
.............. -~

0 -. .~
~

-10 / 1
0 5 10 15 20 25
Distance, x (mm)

Fig. 4 Surface stresses on standard specimen.

ASTM C633-79 standard specimens and elongated specimens Acknowledgments

were found to provide estimates of stresses in these specimens
Funding for this research project was provided by The Ero-
that were in very good agreement with stresses predicted by fi-
sion/Corrosion Research Center at The University of Tulsa and
nite-element models of these specimens. Through the use of fi-
The Oklahoma Center for the Advancement of Science and
nite-element models, it has been shown that the ASTM C633-79
Technology (OCAST).
specimen inherently has a nonuniform stress distribution at the
interface. The peak stress at the interface can be 25% higher than
the average stress assumed by ASTM C633-79. Coating bond References
strengths determined according to this standard could be signifi- 1. ASTM C633-79, "Standard Test Method for Adhesion or Cohesive
cantly lower than actual bond strengths. The elongated speci- Strength of Flame-SprayedCoatings,"Annual Book of ASTM Stand-
men, designed in this study, has a stress distribution that is ard, AmericanSocietyfor Testingand Materials, 1982
uniform to within 0.6% of the average stress assumed by ASTM 2. C.C. Berndt, Tensile Adhesion Testing Methodology for Thermally
C633-79. Laboratory tests using the procedure defined in Sprayed Coatings,./. Mater. Eng., Vol. 12(No. 2), 1990,p 151-158
ASTM C633-79 to determine the tensile strength of an epoxy 3. Y. Shimizu,M. Sato, K. Maeda, and M. Kabayashi,Effect of Test Speci-
showed that the variation in tensile strength can exist from men Size upon AdhesiveStrength of Flame Sprayed Coatings, Ther-
specimen to specimen. However, epoxy tensile strengths using mal Spray Coatings: Properties. Processes and Applications, E
Bemecki, Ed., ASM International,1991,p 257-262
elongated specimens were found to be consistently and signifi-
4. H.S. Ingham, Jr., Adhesion of Flame Sprayed Coatings, Adhesion
cantly higher than tensile strengths using ASTM C633-79 stand-
Measurements of Thin Films, Thick Films, and Bulk Coatings,
ard specimens. The mean tensile strength using the elongated
ASTM STP 640, K.L. Mittal, Ed, American Society for Testingand
specimen was 21% higher than the mean tensile strength using Materials, 1978,p 285-292
the ASTM C633-79 standard specimen. Based on findings from 5. T.N. Rhys-Jones, "Applications of Thermally Sprayed Coating Sys-
the finite-element analyses and from the epoxy tensile tests, it is tems in Aero Engines,"12th Int. Conf. Thermal Spraying, The Weld-
concluded that elongated specimens provide more accurate esti- ing Institute,London, 1989, p 87-99
mates of bond strength than ASTM C633-79 standard speci- 6. R.H. Unger and W.D. Grossklaus, "A Comparison of the Technical
mens. Properties of Arc Sprayed Versus Plasma Sprayed Nickel-5 Alumi-

148~Volume 2(2) June 1993 Journal of Thermal Spray Technology

 Standard and Elongated Specimen
50.0

40.0

30.0
U'J
i ..................ii
CO 20.0

10.0
Standard E!o.n~ated Average

0.0 I I I I I I
0.0 2.0 4.0 6.0 8.0 10.0 12.0
Radius (mm)
Fig. 5 Comparison of axial stresses.

Applied Load: 3000 Ib

60

FE Axial Stress
50
Fore ~orce FE Hoop.Stress

40 Strain Gauges

IX.
30
r
(/)
20
t,=

10

-10 I I L
0 10 20 30 38.1
Distance, x (mm)

Fig. 6 Surface stresses on elongated specimen.

Journal of Thermal Spray Technology Volume 2(2) June 1993---149

 1O0

Elongated Geometry
-ff ASTM Standard
80
Geometry
(/J

-~ 6o
E

40

20 [

L
$1 $2 S3 Sm E1 E2 E3 Em

Fig. 7 Epoxy bond strength tests.

hum," SAE Technical Paper Series, 28th Annual Aerospace/Airline on Test Results, Thermal Spray: International Advances in Coating
Plating & Metal Finishing Forum & Exposition, San Diego, 1992 Technology, C.C. Berndt, Ed., ASM International, 1992, p 911-914
7. R.W. Smith, E. Harzenski, and T. Robisch, "The Structure and Proper- 11. C.C. Berndt, Instrumented Tensile Adhesion Tests on Plasma Sprayed
ties of Thermally Sprayed TiC Particulate Reinforced Steel and Nickel- Thermal Barrier Coatings, J. Mater. Eng., Vol I l (No. 4), 1989, p 275-
Chromium Alloy Powders," 12th Int. Conf. Thermal Spraying, The 282
Welding Institute, Abinton Publishing, London, 1989, p 163-172 12. P. Ostojic and C.C. Berndt, The Variability in Strength of Thermal
8. D.J. Greving and J.R. Shadley, Experimental Evaluations of Thermal Sprayed Coatings, Surf. Coat. Technol., Vo134, 1988, p 43-50
Spray Coatings for Oil Field Equipment Applications, Thermal Spray: 13. S.D. Brown, B.A. Chapman, and G.P. Wirtz, Fracture Kinetics and Me-
International Advances in Coating Technology, C.C. Berndt, Ed., chanical Measurement of Adherence, Thermal Spray Technology,
ASM International, 1992, p 605-610 D.L. Houck, Ed., ASM International, 1989, p 147-157
9. K.M. Howell, Evaluating Bond Strength of Metal Coatings over Con- 14. O. Ambroz and J. Krejcova, Determination of the Adhesive and Cohe-
crete Surfaces, Mater Perform., Vol 3 l(No. 7), 1992, p 29-32 sive Fracture Modes of the Adhesion Tensile Test, Thermal Spray: In-
10. W. Han, E.E Rybicki, and J.R. Shadley, Bond Strength Testing of Ther- ternational Advances in Coating Technology, C.C. Berndt, Ed., ASM
mal Spray Coatings Using ASTM C633-79--Effect of Specimen Size International, 1992, p 921-927

150--Volume 2(2) June 1993 Journal of Thermal Spray Technology