MICROSTRUCTURE, PROCESSING, PERFORMANCE RELATIONSHIPS FOR HIGH
TEMPERATURE COATINGS
Thomas Lillo
Idaho National Laboratory
PO Box 1625, MS 2218 Idaho Falls, ID 83415
Thomas.Lillo@INL.gov, (209) 526-9746
Richard Wright
Idaho National Laboratory
PO Box 1625, MS 2218 Idaho Falls, ID 83415
Richard.Wright@INL.gov, (208) 526-6127
ABSTRACT
High velocity oxy-fuel (HVOF)coating have shown high resistance to corrosion in fossil energy applications and it
is generally accepted that mechanical failure, e.g. cracking or spalling, ultimately will determine coating lifetime.
The use of HVOF thermal spray to apply coatings is one of the most commercially viable and allows the control of
various parameters including powder particle velocity and temperature which influence coating properties, such as
residual stress, bond coat strength and microstructure. Methods of assessing the mechanical durability of coatings
are being developed in order to explore the relationship between HVOF spraying parameters and the mechanical
properties of the coating and coating bond strength. The room temperature mechanical strength, as well as the
resistance of the coating to cracking/spalling during thermal transients, is of considerable importance. Eddy current,
acoustic emission and thermal imaging methods are being developed to detect coating failure during thermal cycling
tests and room temperature tensile tests. Preliminary results on coating failure of HVOF FeAl coatings on carbon
steel, as detected by eddy current measurements during thermal cycling, are presented. The influence of HVOF
coating parameters of iron aluminides - applied to more relevant structural steels, like 316 SS and Grade 91 steel, -
on coating durability will be explored once reliable methods for identification of coating failure have been
developed.
INTRODUCTION
Typically, materials with the high temperature strength and creep resistance required for fossil fuel boilers lack the
necessary corrosion resistance to provide a long service life. Conversely, materials with the needed corrosion
resistance often are not suitable for the high temperature structural requirements or they are difficult to
thermomechanically form into useful shapes. One way of satisfying all the requirements is to apply a coating with
the necessary corrosion resistance to a substrate material that will satisfy the high temperature structural
requirements. The functionality of the system is dependent on the coating being devoid of open porosity, which
would allow corrosive gases to reach the underlying substrate material, and the coating being resistant to cracking or
delamination, which would, again, allow corrosive gases to reach the underlying substrate. Due to these issues,
there has been reluctance to utilize coatings in high temperature, aggressive environments. However, the need to
increase operating temperatures to increase efficiency is driving research to understand the coating process and the
factors that contribute to coating failures. Therefore the current work is focused on understanding the relationship
between coating parameters and coating durability with the goal of optimizing those parameters to produce reliable
coatings with long operating lifetimes.
BACKGROUND
Thermally sprayed coatings as thermal barrier and/or corrosion resistant coatings have long been of interest.
Coating/substrate systems can be chosen to satisfy both the structural and corrosion resistance requirements of
specific applications in high temperature, highly aggressive environments. The high velocity oxygen fuel (HVOF)
thermal spray method can be manipulated to control various characteristics of the applied coating1,2. Figure 1 shows
a schematic of the HVOF technique. A mixture of fuel, for example kerosene and oxygen are fed into a combustion
• Combustion chamber pressure - PC determined by total mass flow of O2 and fuel
Fuel / Oxygen
• Equivalence ratio (phi) - Φ =
( Fuel / Oxygen) Stoich
Figure 1. Schematic representation of the HVOF thermal spray gun.
chamber. The coating in powder form is injected downstream where it is heated and accelerated at a target
(substrate). The powder particles are at least partially melted by the time they impact the substrate where they
spread out over the surface. Particle temperature is controlled, to a large extent, by the oxygen/fuel ratio3 which also
controls whether the atmosphere is oxidizing, neutral or reducing. The pressure in the combustion chamber, as
determined by the feed rate and pressure of the fuel and oxygen, has a strong influence on the particle velocity3,4.
These parameters, along with the CTE of particle and substrate, interact to control the residual stress within the
coating4,5, Figure 2. By active manipulation of these variables, the stress state of the coating can be engineered. The
Residual Stress = Quench + CTE + Peening
- Substrate temp. - Coating CTE - Particle hardness
- Particle temp. - Substrate CTE - Particle velocity
- CTE of particle - Processing temp. - Particle mass
NEUTRAL
TENSILE
200
0
Residual Stress (MPa)
-200
-400
-600
COMPRESSIVE
Stress at coating-substrate interface
Average stress in coating
-800
-1000
350 400 450 500 550 600 650 700
Average Particle Velocity (m/s)
Figure 2. HVOF parameters can be used to control residual stress in the coating.
substrate surface condition, coating strength and the residual stress ultimately control the durability of the coating –
whether it cracks, delaminates and/or spalls. Since the corrosion resistance of coating materials is only slightly
lower than that of wrought material6, the failure of the coating/substrate system typically is attributable to the
mechanical failure of the coating. Elucidation of the interactions of the various HVOF system and materials
parameters is critical in developing high integrity coatings.
TECHNICAL APPROACH
This work focuses on studying the relationship between HVOF processing parameters and the mechanical durability
of the resulting coating. Unfortunately, characterization of the coating durability is difficult since it is not an
inherent materials property and can vary with substrate preparation and HVOF process parameters. The mechanical
stresses and strains under service conditions can be difficult to characterize and simulate under laboratory
conditions. Therefore, characterization of durability is being limited to room temperature cracking behavior of
coatings applied to round tensile bars and the cracking behavior of coating/substrate systems during rapid thermal
cycling. Under both conditions it is necessary to identify when cracking occurs – preferably detecting the first crack
to form in each type of test. To this end, work is focused on developing methods for detecting crack formation in
real time.
The coating material is limited to iron aluminide since this material has shown acceptable corrosion resistance in
previous work6. Carbon steel substrates are being utilized to generate coating/substrate systems with low durability
so that the detection methods and methodology can be more quickly developed. Other, higher temperature
materials, such as grade 91 steel and the nickel-based materials, are more relevant and will be the focus of future
work after the crack detection methods have been refined.
The parameters of interest in this work can be divided into those inherent to the coating and substrate materials and
those that can be varied during HVOF coating deposition. Powder particle velocity, temperature and melted fraction
will be controlled with HVOF parameters which, in conjunction with the CTE of the coating and its mismatch with
the substrate, determine the residual stress, whether it be tensile, neutral or compressive, in the coating. These
factors will be manipulated to generate coatings with different residual stresses which will then be related back to
the durability of the coating.
THERMAL CYCLING
Experimental Set Up
The coating/substrate system will be subjected to thermal cycling over the course of its lifetime. In general there
will be a difference in the CTE between the coating and the substrate materials. This difference can exceed the
strength of the coating and result in crack formation. Furthermore, high operating temperatures may result in
relaxation of the residual stress and even recrystallization which can change the stresses, sign and magnitude, during
thermal cycling. To document this behavior HVOF coatings applied to rods (to eliminate stress risers that would
otherwise occur at sharp corners on a flat dogbone type tensile specimen) have been thermally cycled to failure
(cracking). The rods are nominally 12.7 mm in diameter and 127 mm long. The coating is FeAl and is
approximately 250 microns thick. Figure 3 shows the experimental set up for thermal cycling experiments. The
Gleeble 3500 is used to hold the sample and resistively heat it. Heating rates are currently limited to about 450oC/
minute, although faster or slower rates are easily obtainable. The cooling rate is somewhat slower (~200oC/minute)
but is still fairly high and determined by heat conduction through the water-cooled sample grips. Temperature is
tracked by a type K thermocouple spot welded directly to the coating. A water-cooled eddy current coil surrounds
the sample to detect crack formation during thermal cycling. (Eddy current measurements are highly sensitive to
through thickness cracking which disrupts the eddy currents.)
Thermal
Sprayed
Sample
Water-cooled grip
Eddy current coil
Figure 3. Experimental set up in the Gleeble 3500 for thermal cycling tests.
Results and Discussion
The Gleeble 3500 can be programmed for multiple thermal cycles to a given temperature followed by cooling.
Figure 4 shows information that can be obtained using eddy current coils. The images on the left side in this figure
are generated by scanning an eddy current coil over the surface and plotting the response in shades of gray. Defects
appear bright. The coating appears defect-free in the as-sprayed condition shown in Fig. 4a. During thermal
cycling, however, the eddy current coil is held stationary and designed to interrogate one area of the coating and
As - Sprayed
Real Component (volts)
Eddy Current C-Scan
Absolute Probe
500kHz
b Hold
Heating
Cooling
Real Component (volts)
1st Thermal Cycle (Room Temp. - 700º C)
Eddy Current C-Scan
Absolute Probe 500kHz
Real Component (volts) 2nd Thermal Cycle (Room Temp. - 700º C)
Eddy Current C-Scan
Absolute Probe 500kHz
Figure 4. Image of the coating using an eddy current probe is shown at the left while the response of the in
situ eddy current coil during thermal cycling is shown on the right.
only the output of the coil is of interest. The response of the eddy current coil during thermal cycling is shown for
the first and second thermal cycles on the right side of Fig. 4b and 4c. The shape of the profile is dependent on the
coating (eddy current is a relatively near-surface technique) and the effect of heating/cooling on the magnetic and
electrical properties of the coating, including phase changes. Heating to 700oC causes the coil output to increase
over the first ~2 minutes. The response is steady during a 2 minute hold at 700oC and then decreases as the sample
is cooled back to room temperature. Crack formation is expected to be manifested as an abrupt change in the plot of
coil response with time. However, nothing is evident in the response/time plot even though the eddy current scan,
performed after the thermal cycle, shows numerous cracks (image on the left of Fig. 4b). The response during a
second thermal cycle is very similar to that recorded during the first thermal cycle (note the difference in the vertical
scale between the two plots). The lack of a clear cracking event in these plots is thought to be due to the averaging
effect of the stationary eddy current coil – the coil interrogates an area and the output is an “average” over this area.
A small crack may result in a very small, to the point of being insignificant, change in the output response.
Fabrication of a coil that interrogates a much smaller area is expected to produce a more definitive change in the
output.
A second issue is shown in Fig. 5 in which the sample was thermally cycled up to 800oC for 10 cycles. In this case,
heating resulted in a decreasing output (voltage) up to the temperature where the underlying substrate, a low carbon
steel, begins to transform to austenite. This triggers a dramatic step in the plot of output versus time since electrical
and magnetic properties of austenite are significantly different than the low temperature ferrite phase. The austenite
transforms back to ferrite during cooling and again results in a large step in the output plot. This indicates that the
substrate is having a significant influence on the eddy current coil output – the coil is probing too deep and is
beyond the thickness of the coating which really is the only region of interest. Operation of the coil at higher
frequencies will reduce the depth of penetration and the behavior of the eddy currents and their interaction with
defects (cracks) in the coating will more strongly influence the coil output. (Conversely, if the depth of penetration
is too small then surface roughness of the coating will interfere with crack detection.) Higher frequency operation
may require a new eddy current coil to be designed and fabricated.
Phase Changes
1 cycle
Heating Cooling
Peak-to-valley
Real Component (volts) range
FeAl Coating on Carbon Steel Bar
after 10 Thermal Cycles
Figure 5. FeAl coating on carbon steel rod after thermal cycling 10 times to 800oC. The substrate transform
from ferrite to austenite and back to ferrite again during the thermal cycle.
Figure 5 also shows the peak-to-valley distance for each thermal cycle increases with an increasing number of
cycles. During the first cycle the output voltage ranges from 0.5 to -0.5 volts. After the tenth cycle this peak-to-
valley distance ranges from 0.5 volts to approximately -2.0 volts. Efforts will be made to see if there is a correlation
between the peak-to-valley range and the crack length per unit area and whether it can be used to identify
degradation of the coating. (The other cause for the change in the peak-to-valley range may be microstructural
coarsening.) Once degradation can be reliably detected, much more durable coatings will be studied. It is possible
that a large number of thermal cycles will be needed to induce cracking and automated thermal cycling/data
collection will have to be implemented. The data then can be analyzed offline to determine when cracks started to
form.
Finally Figure 6 shows that delamination of the coating from the substrate may also be significant. However, it
cannot be determined whether the delamination happened before cracking or as a result of cracking. Although
delamination is undesirable, it is not as critical if it occurs without through thickness cracking since there is no path
for the corrosive atmosphere to reach the substrate. Delamination may lead to spalling of the coating, however, and
increased rates of attack on the substrate. Unfortunately, eddy current techniques are not as sensitive to
delaminations in contrast to through thickness cracking. Therefore, delamination detection will be attempted using
thermal imaging. Delaminations are expected to result in “hot” and “cold” spots on the coating surface which
hopefully can be detected in real time using thermal imaging.
Coating
Substrate
Delamination
Figure 6. Optical micrograph of a FeAl coating on carbon steel subjected to 2 thermal cycles to 700oC
showing delamination of the coating from the substrate.
Conclusions
The results of the thermal cycling tests indicate that eddy current measurements may be a viable method for
detecting coating failure by cracking. However, refinement of the eddy current coil design, which increases the
sensitivity to cracking in the coating, is indicated. Also, if the extent of cracking, as indicated by the crack length
per unit area, can be linked to the peak-to-valley range of the eddy current coil output then an automated program
can be implemented to run tests to a high number of thermal cycles on highly durable (crack resistant) coatings.
Finally, thermal imaging will also be implemented in an effort to detect delamination during thermal cycling. Once
the crack and delamination detection methods are developed, comparisons of the durability of HVOF coatings to
commercial coatings, such as weld overlays, will be made.
ROOM TEMPERATURE CRACKING OF COATINGS
Resistance to cracking due to tensile loading is one way of evaluating relative coating performance and can be used
as a screening test for the development of optimum coating systems and parameters. Testing is relatively
straightforward. Tensile specimens of substrate material are coated using HVOF and then loaded in tension until a
crack is formed. Crack formation is quite easily detectable in relatively thick coatings (>150 microns thick) using
acoustic emission techniques, Fig. 7a and 7b. However, the cracking event in thin coatings is not as easily detected
using acoustic emission methods. Spalling of the coating is evident in visual observations, Fig. 7c, and cracking
must be occurring. In future work, eddy current methods will be used in addition to acoustic emission methods to
detect cracking in coating of intermediate thickness (~150 microns) to relatively thin coatings (<100 microns).
a
Coating Thickness Single Crack
410 microns
Tensile Geometry
Coating Thickness Edge cracking
130 microns
Edge cracking
Coating Thickness Spalling
50 microns
Figure 7. Coating behavior during tensile behavior as a function of coating thickness.
Figure 7b and 7c also indicates that cracking in these flat tensile samples initiated at the sharp corner of the tensile
specimen. To eliminate crack initiation at corners, the flat, dogbone style tensile specimens will be replaced with
round tensile samples.
CONCLUSIONS
Crack detection methods, based on eddy current techniques, are being developed to assess the durability of iron
aluminide coatings. Work has focused on detecting cracking during thermal cycling of rods that have been coated
with iron aluminides using HVOF. Crack detection should be possible after slight modifications to the eddy current
coil and the frequency of operation. Coating durability testing during thermal cycling will be automated and the
influence of coating/substrate materials combinations and thermal spray parameters will be determined. Crack
detection by eddy current measurements and acoustic emission methods will be further developed to assess cracking
behavior of coating/substrate combinations at room temperature under an applied tensile load, especially in
relatively thin coating where acoustic emission methods alone have been found to be inadequate to detect crack
formation. The influence of various HVOF parameters on cracking behavior will be determined once suitable crack
detection methods are developed.
REFERENCES
1. Brandt, O.C., J. Thermal Spray Tech., 1995, vol. 4, no.2, pp. 147-152.
2. Totemeier, T.C, Wright, R.N. and Swank, W.D., Intermetallics, 2004, vol. 12, pp. 1335-1344.
3. Totemeier, T.C, Wright, R.N. and Swank, W.D., J. Thermal Spray Tech., 2002, vol. 11, pp. 400-408.
4. Totemeier, T.C, Wright, R.N. and Swank, W.D., Metall.and Mat. Trans A, 2004, vol. 35A, pp. 1807-1814.
5. Clyne, T.W. and Gill, S.C., J. Thermal Spray Tech., 1996, vol. 5, pp. 401-418.
6. Totemeier, T.C and Wright, R.N., in 19th Annual Conference on Fossil Energy Materials, National Energy
Technology Laboratory, Pittsburgh, PA, 2004, available at www.netl.doe.gov.
ACKNOWLEDGMENTS
The authors would like to acknowledge the contributions made by W.D. Swank and D.C. Haggard to prepare the
HVOF thermal spray samples, Dr. D.C. Kunerth for designing the eddy current coil and making the eddy current
measurements and D.E. Clark for setup and operation of the Gleeble 3500 for the thermal cycling experiments. This
report was prepared for the U.S. Department of Energy, Office of Fossil Energy, under DOE Idaho Operations
Office, Contract DE-AC07-05ID14517.
