You are on page 1of 6

Surface & Coatings Technology 200 (2005) 1219 1224

www.elsevier.com/locate/surfcoat

Long exposure steam oxidation testing and mechanical properties


of slurry aluminide coatings for steam turbine components
Alina Aguero a,*, Raul Muelas a, Ana Pastor a, Steve Osgerby b
a

Instituto Nacional de Tecnica Aeroespacial, Area de Materiales Metalicos, Torrejon de Ardoz (Madrid), Spain
b
NPL Materials Centre, National Physical Laboratory, Teddington, Middlesex, W11 0LW United Kingdom
Available online 24 August 2005

Abstract
Important efforts to develop new steels or to protect high creep strength steels in order to allow operation of steam turbines at 650 -C are
being carried out world-wide to increase efficiency. Within the European Project SUPERCOAT (Coatings for Supercritical Steam Cycles),
work has been concentrated in the development of coatings to withstand 50,000 100,000 h of operation at 650 -C under high pressure steam.
Aluminide coatings on ferritic martensitic steels produced by applying an Al slurry followed by a diffusion heat treatment, have shown to be
protective at 650 -C under steam for at least 32,000 h of laboratory steam exposure under atmospheric pressure. Although the as diffused
coatings present through thickness cracks, these do not propagate during exposure to steam or thermal cycling and no new cracks seem to
develop. Moreover, no changes in residual stresses could be observed after thermal cycling. Microstructural characterization of samples at
different periods of exposures has been carried out by SEM-EDS and XRD. The principal mechanism of coating degradation is loss of Al at
the surface due to inwards diffusion. Microhardness as well as Youngs modulus and fracture strength were measured using well established
techniques. The coatings show reasonable ductility (1.6%) when stressed in tension between room temperature and 400 -C which further
increases at higher temperatures providing evidence that the coatings should withstand the mechanical conditions likely to be encountered in
service.
D 2005 Elsevier B.V. All rights reserved.
Keywords: Steam oxidation; Coatings; Aluminide; Slurry; Ferritic martensitic steels; Fracture strength; Youngs modulus

1. Introduction
Future power generation steam turbines are being
designed to have higher efficiencies and to meet stringent
environmental regulations, ensuring plant reliability, availability and maintainability without compromising cost.
Large steam power plants today achieve roughly 45%.
Higher efficiencies can be achieved by higher temperatures
and therefore, the operating temperature is expected to rise
from 550 -C to 650 -C. As part of the European COST
action 522, completed in October 2003, new alloy development activities have been very successful in improving
creep strength, generally achieved through lowering the
chromium content. However, a negative consequence has

* Corresponding author. Tel.: +34 915202561; fax: +34 915201381.


E-mail address: agueroba@inta.es (A. Aguero).
0257-8972/$ - see front matter D 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.surfcoat.2005.07.080

been the worsening of the resistance to steam oxidation [1].


In parallel, a number of commercially available coatings
were explored for steam oxidation protection. These
included materials known to have good high temperature
oxidation resistance and deposited by techniques that can be
employed to coat large steam turbine components either at
the plant or at their location of manufacture, and also taking
into consideration economic aspects. Promising results were
obtained both at the laboratory scale as well as at field
testing [2 4].
Among the tested coatings, diffusion aluminides applied
on P92 are very protective. These coatings are deposited by
applying a commercially available Al slurry followed by a
diffusion heat treatment at 700 -C, and on exposure to steam
the coatings develop a thin protective a-Al2O3. Other
aluminides have recently been produced on ferritic steels
by CVD and pack cementation. CVD aluminides produced
on 9% Cr steels at temperatures higher than 800 -C differ in

1220

A. Aguero et al. / Surface & Coatings Technology 200 (2005) 1219 1224

the initial microstructure as lower Al content is present in


the coating due to a higher degree of coating substrate
interdiffusion [5,6]. Moreover, the deposition temperature is
not realistic for ferritic steels which should not be heated
above 740 -C. Other low temperature MOCVD deposited
aluminides have recently been produced, with the same
microstructure of the slurry deposited aluminide coatings
[7]. Moreover, pack cementation coatings deposited at 700
-C have shown a similar microstructure, and also exhibit
excellent oxidation resistance in pure steam [8].
The results presented in this paper cover long term
laboratory steam oxidation testing results of slurry aluminide
coatings, as well as their interdiffusion behaviour. In addition,
some physical and mechanical properties of as coated
specimens were measured at ambient and elevated temperature employing a range of techniques. This work has been
carried out within the scope of the already mentioned COST
action 522, and a project funded by the European Commission, entitled Coatings for Supercritical Steam Cycles
SUPERCOAT (2002 2005), in which several European research institutions and industries are currently participating.

2. Experimental details
2.1. Substrate material
P92 (C: 0.1, Mn: 0.5, Si: 0.03, Cr: 8.8, Ni: 0.06, Mo: 0.4,
W: 1.8, V: 0.20 wt.%, Fe: bal) was obtained from Nippon
Steel Corporation. Specimens (20  10  3 mm) were
ground (Struers 1200) and vapor degreased prior to coating.
The Al slurry was obtained from commercial sources.
2.2. Coating deposition
The coatings were deposited at INTA by applying the Al
slurry by brush followed by a curing heat treatment at 350
-C for 30 min under air. Diffusion heat treatment was
performed under argon flow or vacuum at 700 -C for 10 h.
2.3. Microstructural characterization
The coated specimens were characterized by XRD
(PHILIPS PW 3710) and by optical and electron microscopy (JEOL JSM-6400 equipped with a Oxford EDS
microanalyzer) and microprobe analysis (JEOL JXA-8900
equipped with an Oxford microanalyser) of metallographically polished cross sections before and after exposure.

2.4.2. Youngs modulus


Youngs modulus was measured at NPL using impact
excitation [9]. Measurements were carried out on rectangular beams, 100  10  5 mm, prior to coating and after
slurry aluminide coating had been applied to the major
faces. The beam is supported at two points approximately
25% and 75% along its length and impacted by dropping a
small ceramic ball onto the center of the top face. The
resonant frequency is recorded via a microphone and
spectrum analyser. The modulus is a function of the
resonant frequency, and the mass and dimensions of the
substrate and coating.
2.4.3. Fracture strength
Fracture strength of the coating in tension was measured
by flexural testing at NPL. Specimens with dimensions,
50  10  3 mm, were coated in the center portion of one of
the major faces. The specimens were strained under 4-pt
bending. Cracking was detected using acoustic emission
(AE) via a waveguide, welded to the specimen.
2.4.4. Thermal cycling
Thermal cycling exposures were carried out in air at
NPL. The thermal cycle consisted of 1 h at 650 -C, followed
by rapid cooling in an air blast for 10 min. Mass change and
residual stress were monitored using well-established X-ray
diffraction techniques [10] every 50 h and 100 h,
respectively, until the exposures were discontinued at 300
h total time at temperature.
2.5. Steam oxidation laboratory testing
The schematics of the closed loop laboratory rig
employed at INTA is shown elsewhere [3]. Prior to testing,
air is displaced from the testing chamber by means of N2
which is kept flowing while heating up to the test temperature (approximately at 600 -C/h). Once the test temperature
is achieved, the N2 flow is cut and the steam flow is
switched on. To carry out weight measurements or to
remove samples, the samples are furnace cooled to about
300 -C under N2 and subsequently removed. The reheat
cycle (from 300 -C to the test temperature) is also carried
out under N2. Samples were removed at different time
intervals for metallographic analysis.

3. Results and discussion


3.1. Slurry aluminide microstructure

2.4. Mechanical characterization


2.4.1. Microhardness
Microhardness measurements were undertaken by taking
the average of three measures with a Vickers microhardness
tester (Leika VMHT MOT) employing an indentation load
of 0.25 N.

Aluminium slurry was applied to P92 by means of a brush


followed by intermediate temperature curing heat treatments for evaporation of the solvent. Fig. 1 shows the
microstructure of the cross section of an as-deposited
aluminium slurry coating after a 700 -C 10 h diffusion heat
treatment under Ar. The diffused coating exhibits 50 60 Am

A. Aguero et al. / Surface & Coatings Technology 200 (2005) 1219 1224

Fe 2Al 5

FeAl
30 m
AlN

10 m

Fig. 1. SEM cross section of a slurry Al coating deposited on P92 after a


700 -C 10 h diffusion heat treatment under Ar.

layer composed by a thick outer layer of Fe2Al5 (Al: 49, Fe:


47, Cr: 4 wt.%) with precipitates rich in Cr (Al 35 wt.%, Cr:
28 wt.%, Fe 37 wt.%), and a thinner FeAl layer (Al: 28, Fe:
67, Cr: 5 wt.%) as characterized by electron diffraction.
Beneath this layer, a 4 5 Am interdiffusion zone with AlN
acicular precipitates is also present. These precipitates result
from N present in the substrate (0.046 wt.%) and not from
inward diffusion (from air), as they are not present in substrates with a N content lower than 0.01 wt.% [4]. The surface
of the coating is very rough as opposed to that of similar
coatings deposited by CVD [5] or pack cementation [8].
The diffused aluminide coating exhibits through thickness stress relieving cracks, perpendicular to the surface,
present already after the initial heat treatment as seen in Fig.
1, likely due to mistmatch between thermal expansion
coefficients of the coating phases and the substrate. The
cracks do not propagate into the substrate and no evidence
of coating substrate spallation has been observed.
3.2. Steam oxidation testing
The diffused slurry aluminide coating exhibits excellent
steam oxidation resistance for a period of up to 32,000 h at
650 -C (test still ongoing) as shown in Fig. 2, where the

weight variation of both coated and uncoated P92 is shown


as a function of time. The initial weight loss of the coated
specimen is due to spalling of residual undiffused slurry [2].
P92 experiences a large increase in weight due to formation
of a thick oxide scale composed of an outer Fe3O4 layer and
inner Fe, Cr spinnel [3,4]. After 15,000 h, spallation of said
oxides begins as shown by the exhibited weight loss.
Fig. 3 shows the Al content at the coatings surface as a
function of the exposure time along with some cross section
micrographs of the diffused aluminized of the corresponding
coatings. During the initial 3000 h the FeAl inner layer
increases in thickness to the expense of Fe2Al5 and at later
stages, the interdiffusion process continues as FeAl transforms into Fe3Al and/or aFe as the Al content at the
coatings surface falls below 20 wt.%. It is noteworthy that
the initially formed cracks do not propagate into the
substrate nor become sites of preferential oxidation attack.
After 30,000 h of exposure the Al concentration near the
coating surface is 8 wt.%, still high enough to sustain a
protective a-Al2O3 scale (as confirmed by XRD).
As Al diffuses inwards the AlN precipitation front
advances into the substrate reaching a thickness of 100
Am after 30,000 h as seen in the attacked cross section
micrographs presented in Fig. 3. The main cause of
degradation of these coatings is therefore the loss of Al
available at the coating surface due to inwards diffusion.
This has been confirmed by the results obtained while
heating coated specimens to 650 -C in air or in Ar
atmosphere, as the coating microstructure evolution is the
same as that of specimens exposed to steam up to at least
5000 h. No, or very little, Al is lost due to spallation of the
protective oxide scale or to formation of volatile species. As
a result of coating substrate interdiffusion, Kirkendall
porosity develops after long exposure (Fig. 3). After
30,000 h the degree of porosity is high but nevertheless,
no evidence of substrate attack was found.
As already mentioned, the surface content of Al is only 8
wt.% after 30,000 h, which is about a third of the expected
life for coated components in a steam environment, but this
concentration may be enough to ensure many more hours of

Weigth Variation (mg/cm2)

57

P-92
47
37

Slurry
Aluminide

27
17
7
-3
0

10000

1221

20000

30000

TIME (h)
Fig. 2. Weight Variation as a function of time for slurry aluminide coated and uncoated P92 exposed to steam at 650 -C.

1222

A. Aguero et al. / Surface & Coatings Technology 200 (2005) 1219 1224

Fe 2 Al 5

30m

FeAl
50

Al surface content ( w % )

45

Fe3Al and/or
-Fe

40
35
30
25

30m

30m

20

30m

15
10

8w%

5
0
0

5000

10000

15000

20000

25000

30000

Time / h
Fig. 3. Surface Al concentration of slurry aluminide coatings exposed to steam at 650 -C as a function of time.

protection. Based on work carried out with high temperature


CVD aluminide coatings on ferritic steels, Y. Zhang and
collaborators conclude that the decrease of the Al surface
content was not significant after extended diffusion times at
temperatures 700 -C [6]. These observations are confirmed by our longer exposure results as after the initial fast
drop in Al content from 50 wt.% to 14 wt.% during the
first 10,000 h, in the 2nd 10,000 h period the Al
concentration only drops by 4% (to 10 wt.%) while during
the last exposed 10,000 h only 2 wt.% Al is lost (to 8 wt.%).
Tests carried out in a by-pass of a steam turbine operating
at 640 -C and 17 MPa at Westfalen show very good
correlation with the laboratory test [3]. Moreover, high
pressure laboratory tests (30 MPa) carried out elsewhere
within the SUPERCOAT project, confirm that pressure does

not appear to have a high influence on the oxidation


behaviour of slurry aluminide coatings.
3.3. Mechanical characterization
3.3.1. Microhardness measurements
HV measurements were carried out on coated specimens
exposed to steam at different intervals. Prior to exposure,
HV of the outer Fe2Al5 layer was of 975 whereas that of the
inner FeAl layer dropped to 540. These values are in
agreement with those measured by Kobayashi and Yakaou
on coatings produced by hot dipping aluminizing [11]. After
30,000 h of exposure the coating has a surface HV of 300
corresponding to Fe3Al (in agreement with an Al content of
8 wt.%).

220
Bar 2
210

Bar 3

Modulus, GPa

Bar 4
200

Bar 5

190
180

170

160
0

100

200

300

400

500

600

700

Temperature, C
Fig. 4. Youngs modulus of 4 samples of a ferritic steel substrate as a function of temperature.

A. Aguero et al. / Surface & Coatings Technology 200 (2005) 1219 1224

1223

7000
6000

First cracking event

AE Energy

5000
4000
3000
2000

Background AE

1000
0
0

0.5

1.5

2.5

3.5

4.5

Strain, %
Fig. 5. AE signals from slurry aluminide coating strained in tension at 400 -C.

E 214:6  0:0233T  9:0  105 T 2

against oxidation, is beneficial in that it inhibits the build


up of stresses in the coating on subsequent thermal cycling.
Consequently crack propagation and spallation of the
coating are unlikely to occur.
Due to the presence of cracks the FYoungs modulus_ of
the coating, becomes an abstract concept. It is possible to
measure local values of modulus for uncracked segments of
the coating but these have little practical relevance.

where T is the temperature measured in -C and E, Youngs


modulus in GPa.
As the coating contained through thickness cracking,
analysis of the impact excitation results for the coated
sample can only be qualitative. The apparent modulus of the
coating was < 20 GPa over the entire temperature range,
which is clearly not representative of dense, adherent
coating: the cracked structure effectively adds mass but
little or no stiffness to the impact excitation specimens
thereby giving the apparent low modulus. The cracked
structure, provided that the coating remains protective

3.3.3. Fracture strength


Examples of the AE signals from the coatings are shown
in Fig. 5. The onset of cracking is determined by the first
high-energy event. When the coating was strained in
tension, the strain to cracking was constant at about 1.6%
from room temperature to 400 -C and then increased with
further increases in temperature as shown in Fig. 6. This
observation correlates with images taken of the surface of
the specimen where it can be seen that the appearance of the
cracks is similar at room temperature and 400 -C having a

3.3.2. Youngs modulus


The Youngs modulus of the ferritic steel substrate as a
function of temperature is shown in Fig. 4. The moduli of
the three coated bars also shown in Fig. 4, were almost
identical and fitted a parabolic relationship with temperature
viz:

3.5

Strain to First Crack, %

20 m

20 m

2.5
2
1.5

40 m

1
0.5
0
0

100

200

300

400

500

600

700

Temperature, C
Fig. 6. Strain to first crack as a function of temperature for slurry aluminide coatings strain in tension.

1224

A. Aguero et al. / Surface & Coatings Technology 200 (2005) 1219 1224

in tension between room temperature and 400 -C which


further increases at higher temperatures. This practical and
low cost coating process would be ideal for steam turbine
components. Low load fatigue testing under steam at 650 -C
is currently underway to determine if the coating cracks will
propagate on loaded components such as rotors and blades.

Acknowledgements

30 m
Fig. 7. SEM cross section of a slurry Al coating deposited on P92 after 300
thermal cycles at 650 -C.

series of parallel cracks with approximately 60 Am spacing.


At 650 -C the cracks are much shorter and widely spaced
indicating the improved ductility of the coating.
3.3.4. Thermal cycling
During thermal cycling the mass change showed an
initial decrease as the undiffused coating spalled during the
initial cycles. The mass of the specimen stayed constant
thereafter indicating that no further spalling of the coating
occurred. Similarly the residual stress in the coating
remained effectively constant at < 0.05 microstrain indicating that thermal stresses were not developing. The
consequence of these two observations is that the coating
did not further crack after repeated thermal cycling as
demonstrated in Fig. 7. This behaviour may be due to the
original stress relieving cracks that develop during the
coating interdiffusion initial heat treatment. In contrast to
these observations, high temperature CVD aluminide coatings initially free of cracks, develop deep reaching cracks
when subjected to thermal cycling [12].

4. Conclusions
Aluminide coatings deposited by slurries have shown
excellent resistance to steam oxidation at 650 -C for a
period exceeding 32,000 h (test still on-going). Although
the heat treated coating presents through thickness cracks,
these do not propagate during exposure to steam or thermal
cycling and no new cracks seem to develop. Moreover, the
coatings show reasonable ductility (1.6%) when stressed

The authors acknowledge the European Commission for


financial support under the 5th framework program
(SUPERCOAT). AA, RM and AP wish to thank all the
members of the Area of Metallic Materials at INTA for their
invaluable technical assistance and F.J. Bolvar at Universidad Complutense de Madrid for his kind contribution in
measuring microhardnesses. SO is grateful to Dr. Louise
Brown, Dr. John Nunn, Mr. Tony Fry and Mr. Jim Banks
(all at NPL) who carried out the experimental work on
measurement of mechanical properties.

References
[1] W.J. Quaddakers, P.J. Ennis, Materials for Advanced Power
Engineering, Universite de Liege, European Commission, Germany,
1998, p. 123.
[2] A. Aguero, J. Garca de Blas, R. Muelas, A. Sanchez, S. Tsipas, Mater.
Sci. Forum 369 372 (2001) 939.
[3] A. Aguero, R. Muelas, R.B. Scarlin, R. Knodler, Materials for
Advanced Power Engineering, Universite de Liege, European Commission, p. 1143.
[4] A. Aguero, R. Muelas, Mater. Sci. Forum 461 464 (2004) 957.
[5] B.A. Pint, Y. Zhang, P.F. Tortorelli, J.A. Haynes, I.G. Wright, Mater.
High Temp. 18 (2001) 185.
[6] Y. Zhang, B.A. Pint, J.A. Haynes, K.M. Cooley, I.G. Wright, 17th
Annual Con. Fossil Energy Materials, U.S. Department of Energy,
April 22 24, Baltimore, Maryland, 2003.
[7] A, Aguero, M. Gutierrez, M. Garca, Materials and Corrosion, in
press.
[8] Z.D. Xiang, P.K. Datta, Surf. Coat. Technol. 184 (2004) 108.
[9] R. Morrell, Traceability calibrating the FGrindosonic_, for impact
excitation modulus measurement, National Physical Laboratory
Report NPL(A), vol. 36, 1996.
[10] M.E. Fitzpatrick, A.T. Fry, P. Holdway, F.A. Kandil, J. Shackleton, L.
Suominen, Determination of Residual Stresses by X-ray Diffraction,
Measurement Good Practice Guide, vol. 52, National Physical
Laboratory (2002 March).
[11] S. Kobayashi, T. Yakou, Mater. Sci. Eng. A338 (2002) 44.
[12] Y. Zhang, B.A. Pint, G.W. Garner, K.M. Cooley, J.A. Haynes, Surf.
Coat. Technol. 188-189 (2004) 35.

You might also like